KSH(1)

📦

version 1.0.10 · 4 versions

1.0.10 · debian trixie, ubuntu questing, ubuntu resolute, ubuntu stonking (latest)
1.0.0~beta.2 · ubuntu jammy
1.0.8 · ubuntu noble
1.0.4 · debian bookworm

, in debian trixie, ubuntu questing, ubuntu resolute, ubuntu stonking

🌸 NAME 🌸

ksh, rksh — KornShell, a standard/restricted command and programming language ✨

okay so bestie, welcome to KSH(1) from the General Commands Manual 💅 this is the whole vibe for ksh and rksh, and we're gonna get through it together, no cap.

✨ SYNOPSIS ✨

ksh [ ±abcefhiklmnprstuvxBCDEGH ] [ ±o option ] ... [ - ] [ arg ... ]
rksh [ ±abcefhiklmnpstuvxBCDEGH ] [ ±o option ] ... [ - ] [ arg ... ]

💖 DESCRIPTION 💖

Ksh is a command and programming language that executes commands read from a terminal or a file. Rksh is a restricted version of the command interpreter ksh; it is used to set up login names and execution environments whose capabilities are more controlled than those of the standard shell. See Invocation below for the meaning of arguments to the shell.

🧠 Definitions. 🧠

so a metacharacter is literally one of these little guys:

;   &   (   )   |   <   >   new-line   space   tab

a blank? that's just a tab or a space, periodt. an identifier is a sequence of letters, digits, or underscores that starts with a letter or underscore. Identifiers are the building blocks of variable names. a vname is one or more identifiers separated by a ., and it can optionally have a .. in front. Vnames are used as function and variable names. a word is a sequence of characters from the character set your current locale defines, but non-quoted metacharacters are not invited to the party.

A command is just a sequence of characters in the shell language syntax. The shell reads each command and does the thing either directly or by calling separate utilities. A built-in command is one the shell handles itself, no separate process needed. Some commands are built in purely for convenience and are not documented here. Built-ins that cause side effects in the shell environment, and built-ins that are found before a path search happens (peep Execution below), are documented here. For historical reasons, some of these built-ins act different from other built-ins, and they're called special built-ins.

👑 Commands. 👑

A simple-command is either a list of variable assignments (see Variable Assignments below) or a sequence of blank separated words that can have a list of variable assignments in front of it (see Environment below). The first word is the name of the command to run. Except as noted below, the rest of the words get passed as arguments to that invoked command. The command name is passed as argument 0 (see exec(2)). The value of a simple-command is its exit status: 0-255 if it terminates normally, and 256+signum if it terminates abnormally (you can get the name of the signal that matches that exit status using the -l option of the kill built-in utility).

A pipeline is one or more commands separated by |. The standard output of every command except the last one gets connected by a socketpair(2) or, if the posix shell option is on, by a pipe(2) to the standard input of the next command. Each command except the last runs asynchronously in a subshell (see Subshells below). If the monitor or pipefail option is on, or if the pipeline has the reserved word time in front of it, then the shell waits for all the component commands in the pipeline to terminate; otherwise, the shell only waits for the last component command. The exit status of a pipeline is the exit status of its last component command, unless the pipefail option is enabled. Every pipeline can have the reserved word ! in front of it, which makes the pipeline's exit status 0 if the last command's exit status is non-zero, and 1 if the last command's exit status is 0.

A list is one or more pipelines separated by ;, &, |&, &&, or ||, and it can optionally end with ;, &, or |&. Out of those five symbols, ;, &, and |& all have equal precedence, which is lower than && and ||. The symbols && and || also have equal precedence. A semicolon (;) means run the preceding pipeline sequentially; an ampersand (&) means run the preceding pipeline asynchronously (i.e., the shell does not wait for that pipeline to finish). The symbol |& means run the preceding pipeline asynchronously with a two-way pipe established to the parent shell; the standard input and output of the spawned pipeline can be written to and read from by the parent shell by applying the redirection operators <& and >& with arg p to commands, and by using the -p option of the built-in commands read and print described later. The symbol && (||) means the list after it gets executed only if the preceding pipeline returns a zero (non-zero) value. One or more new-lines can appear in a list instead of a semicolon, to delimit a command. The first item of the first pipeline of a list that is a simple command not beginning with a redirection, and not occurring within a while, until, or if list, can have a semicolon in front of it. That semicolon is ignored unless the showme option is enabled as described with the set built-in below.

A command is either a simple-command or a compound-command, which is one of the following. Unless otherwise stated, the value returned by a command is that of the last simple-command executed in the command.

for vname [ in word ... ] ;do list ;done — Each time a for command runs, vname gets set to the next word taken from the in word list. If in word ... is omitted, then the for command runs the do list once for each positional parameter that is set, starting from 1 (see Parameter Expansion below). Execution ends when there are no more words in the list.
for (( [expr1] ; [expr2] ; [expr3] )) ;do list ;done — The arithmetic expression expr1 gets evaluated first (see Arithmetic Evaluation below). Then expr2 is repeatedly evaluated until it evaluates to zero, and when it's non-zero, list is executed and expr3 is evaluated. If any expression is omitted, it behaves as if it evaluated to 1.
select vname [ in word ... ] ;do list ;done — A select command prints the set of words on standard error (file descriptor 2), each one preceded by a number. If in word ... is omitted, then the positional parameters starting from 1 are used instead (see Parameter Expansion below). The PS3 prompt is printed, and a line is read from standard input. If that line is the number of one of the listed words, then the variable vname gets set to the word that matches that number. If the line is empty, the selection list gets printed again. Otherwise, the variable vname gets set to the empty string. The contents of the line read from standard input is saved in the variable REPLY. The list runs for each selection until a break or end-of-file is encountered. If the REPLY variable is set to the empty string by the execution of list, then the selection list is printed before displaying the PS3 prompt for the next selection.
case word in [ [(]pattern [ | pattern ] ... ) list ;; ] ... esac — A case command runs the list associated with the first pattern that matches word. The patterns look the same as the ones used for pathname expansion (see Pathname Expansion below). The ;; operator makes case execution terminate. If ;& is used instead of ;;, then the next subsequent list, if there is one, gets executed.
if list ;then list [ ;elif list ;then list ] ... [ ;else list ] ;fi — The list after if runs, and if it returns a zero exit status, the list after the first then runs. Otherwise, the list after elif runs, and if its value is zero, the list after the next then runs. If each successive elif list fails, the else list runs. If the if list has a non-zero exit status and there is no else list, then the if command returns a zero exit status.
while list ;do list ;done until list ;do list ;done — A while command repeatedly runs the while list, and if the exit status of the last command in that list is zero, it runs the do list; otherwise the loop terminates. If no commands in the do list are executed, then the while command returns a zero exit status. until can be used instead of while to negate the loop termination test.
while inputredirection ;do list ;done — Filescan loop. This is defined by a lone input redirection following while (see Input/Output below). It's faster than using the read built-in command in a regular while loop. The shell reads lines from the file or stream opened by inputredirection until the end is reached or the loop is broken. For each line read, the command list is executed with the line's contents assigned to the REPLY variable and the line's fields split into the positional parameters (see Field Splitting and Positional Parameters below). Within the list, standard input is redirected to /dev/null. If the posix compatibility shell option is on, this loop type is disabled and inputredirection is processed like a lone redirection in any other context.
((expression)) — The expression is evaluated using the rules for arithmetic evaluation described below. If the value of the arithmetic expression is non-zero, the exit status is 0; otherwise the exit status is 1.
(list) — Execute list in a subshell (see Subshells below). Note that if two adjacent open parentheses are needed for nesting, a space must be inserted to avoid evaluation as an arithmetic command as described above.
{ list;} — list is simply executed. Note that unlike the metacharacters ( and ), { and } are reserved words and must occur at the beginning of a line or after a ; in order to be recognized.
[[ expression ]] — Evaluates expression and returns a zero exit status when expression is true. See Conditional Expressions below for a description of expression.

function varname { list ;} [ redirection ... ]
varname () compound-command [ redirection ... ]

okay so bestie, this is how you define a function that's referenced by varname. 💖 A function whose varname contains a . is called a discipline function and the portion of the varname preceding the last . must refer to an existing variable. The body of the function is the list of commands between { and }. A function defined with the function varname syntax can also be used as an argument to the . special built-in command to get the equivalent behavior as if the varname() syntax were used to define it. (See Functions below.)

namespace identifier { list ;}

Defines or uses the name space identifier and runs the commands in list in this name space. (See Name Spaces below.)

time [ pipeline ]

If pipeline is omitted the user and system time for the current shell and completed child processes is printed on standard error. Otherwise, pipeline is executed and the elapsed time as well as the user and system time are printed on standard error. The TIMEFORMAT variable may be set to a format string that specifies how the timing information should be displayed. See Shell Variables below for a description of the TIMEFORMAT variable.

The following reserved words are recognized as reserved only when they are the first word of a command and are not quoted:

if then else elif fi case esac for while until do done { } function
select time [[ ]] !

✨ Variable Assignments ✨

One or more variable assignments can start a simple command or can be arguments to the typeset, enum, export, or readonly special built-in commands as well as to other declaration commands created as types. The syntax for an assignment is of the form:

varname=word
varname[word]=word

No space is permitted between varname and the = or between = and word.

varname=(assign_list)

No space is permitted between varname and the =. The variable varname is unset before the assignment. An assign_list can be one of the following:

word ... — Indexed array assignment.
[word]=word ... — Associative array assignment. If preceded by typeset -a this will create an indexed array instead.
assignment ... — Compound variable assignment. This creates a compound variable varname with subvariables of the form varname.name, where name is the name portion of assignment. The value of varname will contain all the assignment elements. Additional assignments made to subvariables of varname will also be displayed as part of the value of varname. If no assignments are specified, varname will be a compound variable allowing subsequence child elements to be defined.
typeset [options] assignment ... — Nested variable assignment. Multiple assignments can be specified by separating each of them with a ;. The previous value is unset before the assignment. Other declaration commands such as readonly, enum, and other declaration commands can be used in place of typeset.
. filename — Include the assignment commands contained in filename.

In addition, a += can be used in place of the = to signify adding to or appending to the previous value. When += is applied to an arithmetic type, word is evaluated as an arithmetic expression and added to the current value. When applied to a string variable, the value defined by word is appended to the value. For compound assignments, the previous value is not unset and the new values are appended to the current ones provided that the types are compatible.

The right hand side of a variable assignment undergoes all the expansion listed below except word splitting, brace expansion, and pathname expansion. When the left hand side of an assignment is a compound variable and the right hand side is the name of a compound variable, the compound variable on the right will be copied or appended to the compound variable on the left.

💬 Comments 💬

A word beginning with # causes that word and all the following characters up to a new-line to be ignored.

🔁 Aliasing 🔁

The first word of each command is replaced by the text of an alias if an alias for this word has been defined. An alias name consists of any number of characters excluding metacharacters, quoting characters, file expansion characters, parameter expansion and command substitution characters, the characters / and =. The replacement string can contain any valid shell script including the metacharacters listed above. The first word of each command in the replaced text, other than any that are in the process of being replaced, will be tested for aliases. If the last character of the alias value is a blank then the word following the alias will also be checked for alias substitution. Aliases can be used to redefine built-in commands but cannot be used to redefine the reserved words listed above. Aliases can be created and listed with the alias command and can be removed with the unalias command.

Aliasing is performed when scripts are read, not while they are executed. Therefore, for an alias to take effect, the alias definition command has to be executed before the command which references the alias is read.

The following aliases are automatically preset when the shell is invoked as an interactive shell. Preset aliases can be unset or redefined.

history=′hist -l′
r=′hist -s′

🌸 Tilde Expansion 🌸

After alias substitution is performed, each word is checked to see if it begins with an unquoted ∼. For tilde expansion, word also refers to the word portion of parameter expansion (see Parameter Expansion below). If a word is preceded by a tilde, then it is checked up to a / to see if it matches a user name in the password database (see getpwname(3)). If a match is found, the ∼ and the matched login name are replaced by the login directory of the matched user. If no match is found, the original text is left unchanged. A ∼ by itself, or in front of a /, is replaced by $HOME, unless the HOME variable is unset, in which case the current user's home directory as configured in the operating system is used. A ∼ followed by a + or - is replaced by $PWD or $OLDPWD respectively.

In addition, when expanding a variable assignment (see Variable Assignments above), tilde expansion is attempted when the value of the assignment begins with a ∼, and when a ∼ appears after a :. A : also terminates a user name following a ∼.

The tilde expansion mechanism may be extended or modified by defining one of the discipline functions .sh.tilde.set or .sh.tilde.get (see Functions and Discipline Functions below). If either exists, then upon encountering a tilde word to expand, that function is called with the tilde word assigned to either .sh.value (for the .sh.tilde.set function) or .sh.tilde (for the .sh.tilde.get function). Performing tilde expansion within a discipline function will not recursively call that function, but default tilde expansion remains active, so literal tildes should still be quoted where required. Either function may assign a replacement string to .sh.value. If this value is non-empty and does not start with a ∼, it replaces the default tilde expansion when the function terminates. Otherwise, the tilde expansion is left unchanged.

🐚 Subshells 🐚

A subshell is a separate execution environment that is a complete duplicate of the current shell environment, except for two things: all traps are reset to default except those for signals that are being ignored, and subshells cannot be interactive (i.e., they have no command prompt). Changes made within a subshell do not affect the parent environment and are lost when the subshell exits.

Particular care should be taken not to confuse a subshell with a newly invoked shell that is merely a child process of the current shell, and which (unlike a subshell) starts from scratch in terms of variables and functions and may be interactive. Beware of shell tutorials on the Internet that confuse these.

Subshells cannot be created or invoked using any command. Instead, the following are automatically run in a subshell:

any command or group of commands enclosed in parentheses;
command substitutions of the first and third form (see Command Substitution below);
process substitutions (see Process Substitution below);
all elements of a pipeline except the last;
any command executed asynchronously (i.e., in a background process).

Creating processes is expensive, so as a performance optimization, a subshell of a non-interactive shell may share the process of its parent environment. Such a subshell is known as a virtual subshell. Subshells are virtual unless or until something (such as asynchronous execution, or an attempt to set a process limit using the ulimit built-in command, or other implementation- or system-defined requirements) makes it necessary to fork(2) it into a separate process. Barring any bugs in the shell, virtual subshells should be indistinguishable from real subshells except by their execution speed and their process ID. See the description of the .sh.pid variable below for more information.

✨ Command Substitution ✨

okay bestie so picture this: you got the standard output from a command list, and you wanna shove it into a word. You can do that with parentheses preceded by a dollar sign — that's $(list) — or with a brace group preceded by a dollar sign — that's ${ list;} — or with a pair of grave accents (``). All of those can be part of a word or the whole word, and any trailing new-lines get yeeted. 💅

In the second case, the { and } are treated as reserved words, so { must be followed by a blank, and } must appear at the beginning of the line or follow a ;.

In the third form — the obsolete one, she's giving legacy — the string between the quotes gets processed for special quoting characters before the command is executed (see Quoting below). 👀

The command substitution $(cat file) can be replaced by the equivalent but faster $(<file). And $(n<#) will expand to the current byte offset for file descriptor n. Not me knowing that. 😭

Except for the second form, the command list runs in a subshell, so no side effects are possible. For the second form, the final } will be recognized as a reserved word after any token.

🌸 Arithmetic Expansion 🌸

An arithmetic expression enclosed in double parentheses preceded by a dollar sign — $(()) — is replaced by the value of the arithmetic expression inside those double parentheses. That's it. That's the whole vibe. ✨

🦋 Process Substitution 🦋

Each command argument of the form <(list) or >(list) will run process list asynchronously, connected to some file in /dev/fd if that directory exists, or else a fifo in a temporary directory. The name of this file becomes the argument to the command. 🫶

If you use the form with >, then writing on this file will provide input for list.
If you use <, then the file passed as an argument will contain the output of the list process.

For example:

paste <(cut -f1 file1) <(cut -f3 file2) | tee >(process1)
>(process2)

This cuts fields 1 and 3 from the files file1 and file2 respectively, pastes the results together, and sends it to the processes process1 and process2, as well as putting it onto the standard output. Slay. 💖

Note that the file, which is passed as an argument to the command, is a UNIX pipe(2), so programs that expect to lseek(2) on the file will not work. Not me warning you. 💀

Process substitution of the form <(list) can also be used with the < redirection operator, which causes the output of list to be standard input or the input for whatever file descriptor is specified. Iconic.

💜 Parameter Expansion 💜

A parameter is a variable, one or more digits, or any of the characters *, @, #, ?, -, $, and !. A variable is denoted by a vname. To create a variable whose vname contains a ., a variable whose vname consists of everything before the last . must already exist. A variable has a value and zero or more attributes. Variables can be assigned values and attributes by using the typeset special built-in command. The attributes supported by the shell are described later with the typeset special built-in command. iykyk.

The shell supports both indexed and associative arrays. An element of an array variable is referenced by a subscript. A subscript for an indexed array is denoted by an arithmetic expression (see Arithmetic Evaluation below) between a [ and a ]. To assign values to an indexed array, use vname=(value ...) or set -A vname value .... The value of all non-negative subscripts must be in the range of 0 through 4,194,303. A negative subscript is treated as an offset from the maximum current index +1 so that -1 refers to the last element. Indexed arrays can be declared with the -a option to typeset. Indexed arrays need not be declared. Any reference to a variable with a valid subscript is legal, and an array will be created if necessary. No cap. 🧚

An associative array is created with the -A option to typeset. A subscript for an associative array is denoted by a string enclosed between [ and ].

Referencing any array without a subscript is equivalent to referencing the array with subscript 0.

The value of a variable may be assigned by writing:

vname=value [ vname=value ] ...

vname[subscript]=value [ vname[subscript]=value ] ...

Note that no space is allowed before or after the =. Periodt.

Attributes assigned by the typeset special built-in command apply to all elements of the array. An array element can be a simple variable, a compound variable or an array variable. An element of an indexed array can be either an indexed array or an associative array. An element of an associative array can also be either. To refer to an array element that is part of an array element, concatenate the subscript in brackets. For example, to refer to the foobar element of an associative array that is defined as the third element of the indexed array, use ${vname[3][foobar]}. That's giving nested main character energy. 👑

A nameref is a variable that is a reference to another variable. A nameref is created with the -n attribute of typeset or with the equivalent nameref command. The value of the variable at the time of that command becomes the variable that will be referenced whenever the nameref variable is used. The name of a nameref cannot contain a .. When a variable or function name contains a ., and the portion of the name up to the first . matches the name of a nameref, the variable referred to is obtained by replacing the nameref portion with the name of the variable referenced by the nameref. If a nameref is used as the index of a for loop, a name reference is established for each item in the list.

A nameref provides a convenient way to refer to the variable inside a function whose name is passed as an argument to a function. For example, if the name of a variable is passed as the first argument to a function, the command typeset -n var=$1 (a.k.a. nameref var=$1) inside the function causes references and assignments to var to be references and assignments to the variable whose name has been passed to the function. Note that, for this to work, the positional parameter must be assigned directly to the nameref as part of the declaration command, as in the example above; only that idiom can allow one function to access a local variable of another. For instance, typeset -n var; var=$1 won't cross that barrier, nor will typeset foo=$1; typeset -n var=foo. The way that's so specific. 😭

If any of the floating point attributes, -E, -F, or -X, or the integer attribute, -i, is set for vname, then the value is subject to arithmetic evaluation as described below.

Positional parameters, parameters denoted by a number, may be assigned values with the set special built-in command. Parameter $0 is set from argument zero when the shell is invoked.

The character $ is used to introduce substitutable parameters.

Now the parameter expansions, bestie. Buckle up. 💅

${parameter} — The shell reads all the characters from ${ to the matching } as part of the same word even if it contains braces or metacharacters. The value, if any, of the parameter is substituted. The braces are required when parameter is followed by a letter, digit, or underscore that is not to be interpreted as part of its name, when the variable name contains a .. The braces are also required when a variable is subscripted unless it is part of an Arithmetic Expression or a Conditional Expression. If parameter is one or more digits then it is a positional parameter. A positional parameter of more than one digit must be enclosed in braces. If parameter is * or @, then all the positional parameters, starting with $1, are substituted (separated by a field separator character). If an array vname with last subscript *, @, or for indexed arrays of the form sub1 .. sub2 is used, then the value for each of the elements between sub1 and sub2 inclusive (or all elements for * and @) is substituted, separated by the first character of the value of IFS.
${#parameter} — If parameter is * or @, the number of positional parameters is substituted. Otherwise, the length of the value of the parameter is substituted.
${#vname[*]} ${#vname[@]} — The number of elements in the array vname is substituted.
${@vname} — Expands to the type name (see Type Variables below) or attributes of the variable referred to by vname.
${!vname} — Expands to the name of the variable referred to by vname. This will be vname except when vname is a name reference.
${!vname[subscript]} — Expands to name of the subscript unless subscript is *, @, or of the form sub1 .. sub2. When subscript is *, the list of array subscripts for vname is generated. For a variable that is not an array, the value is 0 if the variable is set. Otherwise it is the empty string. When subscript is @, same as above, except that when used in double quotes, each array subscript yields a separate argument. When subscript is of the form sub1 .. sub2 it expands to the list of subscripts between sub1 and sub2 inclusive using the same quoting rules as @.
${!prefix@} ${!prefix*} — These both expand to the names of the variables whose names begin with prefix. The expansions otherwise work like $@ and $*, respectively (see under Quoting below).
${parameter:-word} — If parameter is set and has a non-empty value, then substitute its value; otherwise substitute word.
${parameter:=word} — If parameter is not set or has the empty string value, then set it to word; the value of the parameter is then substituted. Positional parameters may not be assigned to in this way.
${parameter:?word} — If parameter is set and has a non-empty value, then substitute its value; otherwise, print word and exit from the shell (if not interactive). If word is omitted then a standard message is printed.
${parameter:+word} — If parameter is set and has a non-empty value then substitute word; otherwise substitute the empty string.

In the above, word is not evaluated unless it is to be used as the substituted string, so that, in the following example, pwd is executed only if d is not set or has the empty string value:

print ${d:-$(pwd)}

If the colon (:) is omitted from the above expressions, then the shell only checks whether parameter is set or not.

That's the whole parameter expansion situation. Obsessed. 🎀

${parameter:offset:length} and ${parameter:offset} — okay so this is the slice it up expansion 💅. It expands to the portion of parameter's value that starts at the character position you get by expanding offset as an arithmetic expression (counting from 0, bestie, not from 1, we're not animals), and it runs for however many characters the arithmetic expression in length says. In the second form there's no length, so you just get the remainder of the value. A negative offset counts backwards from the end of parameter.

Heads up: you need one or more blanks in front of a minus sign, or the shell will read the operator as :- and your whole vibe collapses. 😭

If parameter is * or @, or it's an array name indexed by * or @, then offset and length refer to the array index and the number of elements respectively. A negative offset is taken relative to one greater than the highest subscript for indexed arrays. For associative arrays, the order is unspecified — so like, don't count on it, fr fr.

${parameter#pattern} and ${parameter##pattern} — the trim it off the front girls 🎀. If the shell pattern matches the beginning of parameter's value, then this expansion is parameter's value with the matched portion deleted; if it doesn't match, you just get parameter's value back, no notes. In the first form the smallest matching pattern gets deleted, and in the second form the largest matching pattern gets deleted. When parameter is @, *, or an array variable with subscript @ or *, the substring operation is applied to each element in turn.
${parameter%pattern} and ${parameter%%pattern} — same energy but from the back 🦋. If the shell pattern matches the end of parameter's value, the expansion is parameter's value with the matched part deleted; otherwise you just get parameter's value. First form deletes the smallest matching pattern, second form deletes the largest matching pattern. And yes, when parameter is @, *, or an array variable with subscript @ or *, the substring operation is applied to each element in turn.
${parameter/pattern/string}, ${parameter//pattern/string}, ${parameter/#pattern/string}, ${parameter/%pattern/string} — the find-and-replace squad, and they ate 🔥. This expands parameter and replaces the longest match of pattern with the given string. Every occurrence of \n in string gets replaced by the portion of parameter that matches the n-th subpattern. Here's the breakdown:
- First form: only the first occurrence of pattern is replaced.
- Second form: every match of pattern is replaced by string.
- Third form: the pattern match is restricted to the beginning of the string.
- Fourth form: the pattern match is restricted to the end of the string.
- In the first and second forms, an empty pattern never matches. In the third and fourth forms, an empty pattern matches the beginning or the end of the string, respectively.
- When string is empty, the pattern gets deleted and the / in front of string may be omitted.
- When parameter is @, *, or an array variable with subscript @ or *, the substitution operation is applied to each element in turn. In this case, the string portion of word will be re-evaluated for each element. 👑

🌸 Shell Variables 🌸

The following parameters are automatically set by the shell, no cap, you don't have to do a thing:

# — the number of positional parameters, in decimal.
- — the options supplied to the shell on invocation, or by the set command.
? — the exit status returned by the last executed command. Its meaning depends on the command or function that defines it, but there are conventions that other commands often depend on: zero typically means 'success' or 'true', one typically means 'non-success' or 'false', and a value greater than one typically indicates some kind of error. Only the 8 least significant bits of $? (values 0 to 255) are preserved when the exit status is passed on to a parent process, but within the same (sub)shell environment it's a signed integer value with a range of possible values as shown by the commands getconf INT_MIN and getconf INT_MAX. Shell functions that run in the current environment may return status values in this range.
$ — the process ID of the main shell process. Note that this value will not change in a subshell, even if the subshell runs in a different process. See also .sh.pid.
_ — okay this one has lore, sit down bestie 🫶. Initially, the value of _ is an absolute pathname of the shell or script being executed as passed in the environment. Subsequently it is assigned the last argument of the previous command. This parameter is not set for commands which are asynchronous. It's also used to hold the name of the matching MAIL file when checking for mail. While defining a compound variable or a type, _ is initialized as a reference to the compound variable or type. When a discipline function is invoked, _ is initialized as a reference to the variable associated with the call to this function. And finally, when _ is used as the name of the first variable of a type definition, the new type is derived from the type of the first variable. (See Type Variables below.)
! — the process ID of the last background command invoked, or the most recent job put in the background with the bg built-in command.
.sh.command — when processing a DEBUG trap, this variable contains the current command line that is about to run. The value is in the same format as the output generated by the xtrace option (minus the preceding PS4 prompt).
.sh.edchar — this variable contains the value of the keyboard character (or sequence of characters if the first character is ESC, ASCII 033) that has been entered when processing a KEYBD trap (see Key Bindings below). If the value is changed as part of the trap action, then the new value replaces the key (or key sequence) that caused the trap.
.sh.edcol — the character position of the cursor at the time of the most recent KEYBD trap.
.sh.edmode — upon executing a KEYBD trap action, the value of this variable is set to the ESC control character if the shell is in vi input mode (see Vi Editing Mode below), or to the empty string value otherwise.
.sh.edtext — the characters in the input buffer at the time of the most recent KEYBD trap. The variable is unset when not processing a KEYBD trap.
.sh.file — the pathname of the file that contains the current command.
.sh.fun — the name of the current function that is being executed.
.sh.level — set to the current call depth of functions and dot scripts. Normally this variable is read-only, but while executing a DEBUG trap, its value may be changed to switch the current function scope to that of the specified level for the duration of the trap run, which makes it possible to access a parent scope for debugging purposes. When trap execution ends, the variable and the scope are restored. It is an error to assign a value lower than 0 (the global scope) or higher than the current call depth.
.sh.lineno — set during a DEBUG trap to the line number for the caller of each function.
.sh.match — whenever a match is found in a pattern matching operation using either the [[ compound command (see Conditional Expressions below) or the expansions ${parameter#pattern}, ${parameter%pattern}, ${parameter/pattern/string}, or one of their variants (see Parameter Expansion above), the match and its subpattern matches are stored in this indexed array, overwriting its previous values. The 0-th element stores the complete match and the i-th element stores the i-th submatch. For //, the array is two-dimensional, with the first subscript indicating the most recent match and subpattern match, and the second subscript indicating which match, with 0 representing the first match. If no match is found, .sh.match is not reset or modified. Note that even matching operations performed on the .sh.match variable itself will overwrite it upon finding a match. 💀
.sh.math — used for defining arithmetic functions (see Arithmetic Evaluation below) and stores the list of user defined arithmetic functions.
.sh.name — set to the name of the variable at the time that a discipline function is invoked.
.sh.subscript — set to the name subscript of the variable at the time that a discipline function is invoked.
.sh.subshell — the current depth for subshells and command substitution.
.sh.pid — set to the process ID of the current shell process. Unlike $$, this is updated in a subshell when it forks into a new process. Note that a virtual subshell may have to fork mid-execution due to various system- and implementation-dependent requirements, so the value should not be counted on to remain the same from one command to the next. If a persistent process ID is required for a subshell, it must be ensured it is running in its own process first. Any attempt to set a process limit using the ulimit built-in command, such as ulimit -t unlimited 2>/dev/null, is a reliable way to make a subshell fork if it hasn't already. ✨
.sh.ppid — set to the process ID of the parent of the current shell process 💅 and here's the ✨ twist that makes it different from $PPID: this one does get updated in a subshell when it forks into a new process. The same note as for .sh.pid applies, bestie 🫶
.sh.value — set to the value of the variable at the time that the set or append discipline function is invoked 👑. When a user defined arithmetic function is invoked, the value of .sh.value is saved and .sh.value is set to long double precision floating point, then .sh.value is restored when the function returns 🔁
.sh.version — set to a value that identifies the version of this shell ⭐
COLUMNS — the width of the terminal window in character positions 🎀. It gets updated automatically at initialization and on receiving a SIGWINCH signal, and the shell uses the value to define the width of the edit window for the shell edit modes and for printing select lists 📐
KSH_VERSION — a name reference to .sh.version 🪞
LINENO — the current line number within the script or function being executed 🔢
LINES — the height of the terminal window in lines 🌈. Updated automatically at initialization and on receiving a SIGWINCH signal, and the shell uses the value to determine the column length for printing select lists: they get printed vertically until about two thirds of LINES lines are filled 📏
OLDPWD — the previous working directory, set by the cd command 👣
OPTARG — the value of the last option argument processed by the getopts built-in command 🍬
OPTIND — the index of the last option argument processed by the getopts built-in command 🔢
PPID — the process ID of the parent of the main shell process 💜. Note that this value will not change in a subshell, even if the subshell runs in a different process. See also .sh.ppid 👀
PWD — the present working directory, set by the cd command 🏠
RANDOM — every single time this variable is referenced, a pseudorandom integer is generated, uniformly distributed between 0 and 32767 (the 16-bit unsigned integer range) except that the same number is never repeated twice in a row 🎲. The sequence of pseudorandom numbers is reproducible, and you can initialize it to a fixed starting point by assigning a numeric seed value to RANDOM 🌱. Each time a new shell or subshell environment is entered (see Subshells above), the sequence is automatically reset to a different point 🔄
REPLY — this variable is set by the select statement and by the read built-in command when no arguments are supplied 💌
SECONDS — every time this variable is referenced, the number of seconds since shell invocation is returned ⏱️. If this variable is assigned a value, then the value returned upon reference will be the value that was assigned plus the number of seconds since the assignment 🕰️
SHLVL — an integer variable that is incremented and exported each time the shell is invoked 📈. If SHLVL is not in the environment when the shell is invoked, it is set to 1 1️⃣

The following variables are used by the shell 🌈

okay bestie, second squad incoming — these ones the shell actually reads and vibes with, so listen up 🫶

CDPATH — the search path for the cd command 🗺️
EDITOR — if the VISUAL variable is not set, the value of this variable will be checked for certain patterns and the corresponding editing option will be turned on as described with VISUAL below ✍️
ENV — if this variable is set, then parameter expansion, command substitution, and arithmetic expansion are performed on the value to generate the pathname of the script that will be executed when the shell is invoked interactively (see Invocation below) 💫. This file is typically used for alias and function definitions, and the default value is $HOME/.kshrc 🏡. On systems that support a system wide /etc/ksh.kshrc initialization file, if the filename generated by the expansion of ENV begins with /./ or ././ the system wide initialization file will not be executed 🚫
FCEDIT — obsolete name for the default editor name for the hist command 💀. FCEDIT is not used when HISTEDIT is set
FIGNORE — a pattern that defines the set of filenames that will be ignored when performing filename matching 🙈
FPATH — the search path for function definitions 🔍. The directories in this path are searched for a file with the same name as the function or command when a function with the -u attribute is referenced and when a command is not found. If an executable file with the name of that command is found, then it is read and executed in the current environment 🏃‍♀️. Unlike PATH, the current directory must be represented explicitly by . rather than by adjacent : characters or a beginning or ending : ⚠️
histchars — this variable can be used to specify up to three ASCII characters that control history expansion (see History Expansion below) 🕰️. The first (default: !) signals the start of a history expansion. The second (default: ^) is used for short-form substitutions. The third (default: #), when found as the first character of a word, causes history expansion to be skipped for the rest of the words on the line 🙅‍♀️. Multi-byte characters (e.g. UTF-8) are not supported and produce undefined results 💀
HISTCMD — number of the current command in the history file 📜
HISTEDIT — name for the default editor name for the hist command ✏️
HISTFILE — if this variable is set when the shell is invoked, then the value is the pathname of the file that will be used to store the command history (see Command Re-entry below) 🗂️
HISTSIZE — if this variable is set when the shell is invoked, then the number of previously entered commands that are accessible by this shell will be greater than or equal to this number 👑. The default is 512
HOME — the default argument (home directory) for the cd command 🏠
IFS — internal field separators, normally space, tab, and new-line, that are used to separate the results of command substitution or parameter expansion and to separate fields with the built-in command read ✂️. The first character of the IFS variable is used to separate arguments for the "$*" expansion (see Quoting below). Each single occurrence of an IFS character in the string to be split that is not in the isspace character class, and any adjacent characters in IFS that are in the isspace character class, delimit a field. One or more characters in IFS that belong to the isspace character class delimit a field. In addition, if the same isspace character appears consecutively inside IFS and the posix shell option is not on, this character is treated as if it were not in the isspace class — for example, if IFS consists of two tab characters, then two adjacent tab characters delimit an empty field 😳
JOBMAX — this variable defines the maximum number running background jobs that can run at a time 🏃. When this limit is reached, the shell will wait for a job to complete before starting a new job ⏳
LANG — this variable determines the locale category for any category not specifically selected with a variable starting with LC_ or LANG 🌍
LC_ALL — this variable overrides the value of the LANG variable and any other LC_ variable 👑
LC_COLLATE — this variable determines the locale category for character collation information 🔤
LC_CTYPE — this variable determines the locale category for character handling functions. It determines the character classes for pattern matching (see Pathname Expansion below) 🧬
LC_NUMERIC — this variable determines the locale category for the decimal point character 🔢
MAIL — if this variable is set to the name of a mail file and the MAILPATH variable is not set, then the shell informs the user of arrival of mail in the specified file 📬
MAILCHECK — this variable specifies how often (in seconds) the shell will check for changes in the modification time of any of the files specified by the MAILPATH or MAIL variables ⏲️. The default value is 600 seconds, and when the time has elapsed the shell will check before issuing the next prompt ⏰
MAILPATH — a colon ( : ) separated list of file names 📮. If this variable is set, then the shell informs the user of any modifications to the specified files that have occurred within the last MAILCHECK seconds. Each file name can be followed by a ? and a message that will be printed 💌. The message will undergo parameter expansion, command substitution, and arithmetic expansion with the variable $_ defined as the name of the file that has changed. The default message is you have mail in $_ 📨
PATH — the search path for commands (see Execution below) 🛣️. The user may not change PATH if executing under rksh (except in .profile) ⚠️

PS1

okay so every single time a new command line gets started on an interactive shell, the value of this variable gets expanded — and we're talking backslash escaping, parameter expansion, command substitution, and arithmetic expansion all happening — and whatever falls out of that is literally the primary prompt string for that command line. the default is $ (space included, bestie). the ! character in the primary prompt string gets replaced by the command number (peep Command Re-entry below 👀). two !s in a row will produce a single ! when the prompt string is printed. and heads up: any terminal escape sequences you use in the PS1 prompt therefore need every single instance of ! in them changed to !!. 💅

PS2

secondary prompt string, default is > .

PS3

selection prompt string used inside a select loop, default is #? .

PS4

the value of this variable gets expanded for parameter evaluation, command substitution, and arithmetic expansion, and it precedes each line of an execution trace. by default, PS4 is + . and also, when PS4 is unset, the execution trace prompt is + too. ✨

SHELL

the pathname of the shell is kept in the environment. at invocation, if the basename of this variable is rsh, rksh, or krsh, then the shell becomes restricted. 🚫

TIMEFORMAT

the value of this parameter is used as a format string specifying how the timing information for pipelines prefixed with the time reserved word should be displayed. the % character introduces a format sequence that gets expanded to a time value or other information. the format sequences and their meanings are as follows:

%% — a literal %.
%[p][l]R — the elapsed time in seconds.
%[p][l]U — the number of CPU seconds spent in user mode.
%[p][l]S — the number of CPU seconds spent in system mode.
%P — the CPU percentage, computed as (U + S) / R.

the brackets denote optional portions. the optional p is a digit specifying the precision — the number of fractional digits after a decimal point. a value of 0 causes no decimal point or fraction to be output. at most three places after the decimal point can be displayed; values of p greater than 3 are treated as 3. if p is not specified, the value 3 is used.

the optional l specifies a longer format, including hours if greater than zero, minutes, and seconds of the form HHhMMmSS.FFs. the value of p determines whether or not the fraction is included. seconds are zero-padded unless the posix shell option is on.

all other characters are output without change and a trailing newline is added. if the variable is unset, the default value, $'\nreal\t%2lR\nuser\t%2lU\nsys\t%2lS', is used. if the value is empty, no timing information is displayed. 🙈

TMOUT

terminal read timeout. if set to a value greater than zero, the read built-in command and the select compound command time out after TMOUT seconds when input is from a terminal. an interactive shell will issue a warning and allow for an extra 60 second timeout grace period before terminating if a line is not entered within the prescribed number of seconds while reading from a terminal. (note that the shell can be compiled with a maximum bound for this value which cannot be exceeded. 😭)

VISUAL

the value of this variable is scanned when the shell is invoked and whenever its value is changed; if it's found to match certain patterns, the corresponding line editor (see In-line Editing Options below) is activated. if it matches the pattern *[Vv][Ii]*, the vi option is turned on; else if it matches the pattern *gmacs*, the gmacs option is turned on; else if it matches the pattern *macs*, the emacs option is turned on. if none of the patterns match, emacs is turned on by default upon initializing an interactive shell. if the value is changed by assignment and none of the patterns match, no options are changed. the value of VISUAL overrides the value of EDITOR. 👑

the shell gives default values to PATH, PS1, PS2, PS3, PS4, MAILCHECK, FCEDIT, TMOUT and IFS, while HOME, SHELL, ENV, histchars, and MAIL are not set at all by the shell (although HOME is set by login(1)). on some systems MAIL and SHELL are also set by login(1). 🫶

🌸 Field Splitting 🌸

after parameter expansion and command substitution, the results of substitutions are scanned for the field separator characters (those found in IFS) and split into distinct fields wherever such characters are found. explicit empty fields ("" or '') are retained. implicit empty fields — those resulting from parameters that are unset or have empty string values, or from command substitutions yielding the empty string, and that are not quoted with "" — are removed. 💀

🎀 Brace Expansion 🎀

if the braceexpand (-B) option is set, then each word, as well as any fields resulting from field splitting (see above), gets checked to see if they contain one or more of the brace patterns {*,*}, {l1..l2}, {n1..n2}, {n1..n2%fmt}, {n1..n2..n3}, or {n1..n2..n3%fmt}, where * represents any character, l1,l2 are letters and n1,n2,n3 are signed numbers and fmt is a format specified as used by printf. in each case, fields are created by prepending the characters before the { and appending the characters after the } to each of the strings generated by the characters between the { and }. the resulting fields are checked to see if they have any brace patterns. 🔁

in the first form, a field is created for each string between { and ,, between , and ,, and between , and }. the string represented by * can contain embedded matching { and } without quoting. otherwise, each { and } with * must be quoted.

in the second form, l1 and l2 must both be either upper case or both be lower case characters in the C locale. in this case a field is created for each character from l1 through l2.

in the remaining forms, a field is created for each number starting at n1 and continuing until it reaches n2, incrementing n1 by n3. the values of n1, n2 and n3 are limited to the standard integer range as output by getconf INT_MIN and getconf INT_MAX; the behavior is undefined if this range is exceeded. the cases where n3 is not specified behave as if n3 were 1 if n1<=n2 and -1 otherwise. in forms which specify %fmt, any format flags, widths and precisions can be specified and fmt can end in any of the specifiers cdiouxX. for example, {a,z}{1..5..3%02d}{b..c}x expands to the 8 fields, a01bx, a01cx, a04bx, a04cx, z01bx, z01cx, z04bx and z04cx. 🌈

🦋 Pathname Expansion 🦋

this is also known as globbing or sometimes filename generation. pathname expansion is disabled if the -f a.k.a. --noglob shell option is on. otherwise, if certain special characters are found in a word or in a field resulting from field splitting (see above), then the word or field is regarded as a pattern. each literal word is scanned for the characters *, ?, [, and (, but fields resulting from field splitting are scanned only for the characters *, ?, and [ for compatibility reasons (in which case the ( character is not special and any pattern syntax described below that involves parentheses does not apply).

each file name component that contains a recognized pattern character is replaced with a lexicographically sorted set of names that matches the pattern from that directory. if no file name is found that matches the pattern, then that component of the filename is left unchanged — unless the pattern is prefixed with ~(N), in which case it is removed as described below. the special traversal names . and .. are never matched. if FIGNORE is set, then each file name component that matches the pattern defined by the value of FIGNORE is ignored when generating the matching filenames. if FIGNORE is not set, the character . at the start of each file name component will be ignored unless the first character of the pattern corresponding to this component is the character . itself. note that, for uses of pattern matching other than pathname expansion, the / and . are not treated specially. 👀

* — matches any string, including the empty string. when used for filename expansion, if the globstar option is on, an isolated pattern of two adjacent *s will match all files and zero or more directories and subdirectories. if followed by a / then only directories and subdirectories will match.
? — matches any single character.
[...] — matches any one of the enclosed characters. a pair of characters separated by - matches any character lexically between the pair, inclusive. if the first character following the opening [ is a ! or ^, then any character not enclosed is matched. a - can be included in the character set by putting it as the first or last character.

within [ and ], character classes can be specified with the syntax [:class:], where class is one of the following classes defined in the ANSI C standard (note that word is equivalent to alnum plus the character _):

alnum alpha blank cntrl digit graph lower print punct
space upper word xdigit

within [ and ], an equivalence class can be specified with the syntax [=c=] which matches all characters with the same primary collation weight (as defined by the current locale) as the character c. within [ and ], [.symbol.] matches the collating symbol symbol. ⭐

Okay so a pattern-list is literally just a list of one or more patterns, glued together with either & or |. The & is a "no, ALL of y'all have to match" situation, whereas | is chill and only needs ONE of the patterns to match. 🫶

You can build composite patterns out of one or more of these subpatterns:

?(pattern-list) — optionally matches any one of the given patterns.
*(pattern-list) — matches zero or more occurrences of the given patterns.
+(pattern-list) — matches one or more occurrences of the given patterns.
{n}(pattern-list) — matches n occurrences of the given patterns.
{m,n}(pattern-list) — matches from m to n occurrences of the given patterns. Leave m out and it's treated as 0. Leave n out and it means at least m occurrences get matched.
@(pattern-list) — matches exactly one of the given patterns.
!(pattern-list) — matches anything except one of the given patterns.

Default behavior: each pattern or subpattern grabs the longest string possible that still lets the overall match be the longest. If more than one match is possible, the shell goes with the one that starts closest to the beginning of the string. BUT 💅 — for each of the compound patterns above, you can slide a - in front of the ( to make it use the shortest match to the specified pattern-list instead. Bestie, we love options.

When pattern-list is sitting inside parentheses, the backslash character \ is treated specially — even when it's inside a character class. All ANSI C character escapes are recognized and match the character they say they do. And on top of that, these escape sequences work too:

\d — matches any character in the digit class.
\D — matches any character not in the digit class.
\s — matches any character in the space class.
\S — matches any character not in the space class.
\w — matches any character in the word class.
\W — matches any character not in the word class.

A pattern of the form %(pattern-pair(s)) is a subpattern you use to match nested character expressions. Each pattern-pair is a two-character sequence, and it can't contain & or |. The very first pattern-pair gives the starting and ending characters for the match. Every pattern-pair after that represents the beginning and ending characters of a nested group that gets skipped over when the shell is counting starting and ending character matches. 🌀

Heads up: the behavior is unspecified when the first character of a pattern-pair is alphanumeric — except for the following, which are totally defined: - D — makes the ending character terminate the search for this pattern without finding a match. - E — makes the ending character be interpreted as an escape character. - L — makes the ending character be interpreted as a quote character, causing all characters to be ignored when looking for a match. - Q — makes the ending character be interpreted as a quote character, causing all characters other than any escape character to be ignored when looking for a match.

So like, %({}Q"E\) matches characters starting at { until the matching } is found, and it does NOT count any { or } that's inside a double-quoted string or preceded by the escape character \. Drop the {} and this pattern matches any C language string. 🌈

Each subpattern in a composite pattern gets numbered, starting at 1, based on where its ( shows up in the pattern. The sequence \n, where n is a single digit and \n comes after the nth subpattern, matches the same string as the subpattern itself. Cute, right?

And finally (yes bestie we made it), a pattern can contain subpatterns of the form ∼(options:pattern-list), where either options or :pattern-list can be left out. Unlike the other compound patterns, these subpatterns do NOT get counted in the numbered subpatterns. For options E, F, G, N, P, V, and X below, :pattern-list must be omitted. If options is present, it can consist of one or more of the following:

+ — enable the following options. This is the default.
- — disable the following options.
E — the remainder of the pattern uses extended regular expression syntax, like the -E option of the grep(1) command.
F — the remainder of the pattern uses the fixed pattern syntax of the -F option of the grep(1) command.
G — the remainder of the pattern uses basic regular expression syntax, like the grep(1) command without options.
K — the remainder of the pattern uses shell pattern syntax. This is the default.
N — when it's the first letter and it's used with pathname expansion, and no matches occur, the file pattern expands to the empty string instead of staying unexpanded. Otherwise, it's ignored.
X — the remainder of the pattern uses augmented regular expression syntax, like the -X option of the AT&T AST version of the grep(1) command.
P — the remainder of the pattern uses perl(1) regular expression syntax. Not all perl regular expression syntax is currently implemented.
V — the remainder of the pattern uses System V regular expression syntax.
i — always treat the match as case-insensitive, regardless of the globcasedetect shell option.
g — file the longest match (greedy). This is the default.
l — left-anchor the pattern. This is the default for K style patterns.
r — right-anchor the pattern. This is the default for K style patterns.

If both options and :pattern-list are specified, then the options apply only to pattern-list. Otherwise, those options stay in effect until a later ∼(...) disables them, or until you hit the end of the subpattern that contains the ∼(...). ✨

Quoting 💬

Each of those metacharacters we listed earlier (peep Definitions above, iykyk) means something special to the shell, and it straight up terminates a word unless you quote it. A character can be quoted (i.e. made to stand for itself) by putting a \ in front of it. The pair \new-line gets removed. 🫶

All characters wrapped between a pair of single quote marks (′′) that isn't preceded by a $ are quoted. A single quote cannot appear inside the single quotes — sorry, that's just the rules. And a single quoted string preceded by an unquoted $ gets processed as a C99 string, except for the following:

\0 — causes the remainder of the string to be ignored.
\E — equivalent to the escape character (ASCII 033),
\e — equivalent to the escape character (ASCII 033),
\cx — expands to the character control-x.
\C[.name.] — expands to the collating element name.

Inside double quote marks (""), parameter and command substitution happen, and \ quotes the characters \, `, ", and $. A $ in front of a double quoted string gets ignored in the "C" or "POSIX" locale, and it may cause the string to be replaced by a locale specific string otherwise. The meaning of $* and $@ is identical when they're not quoted, or when they're used as a variable assignment value, or as a file name. BUT when they're used as a command argument, "$*" is equivalent to "$1d$2d...", where d is the first character of the IFS variable, whereas "$@" is equivalent to "$1" "$2" .... 🌸 Inside grave quote marks ( ), `\` quotes the characters `\`, ``, and$. And if the grave quotes are inside double quotes, then ` also quotes the character ".

The special meaning of reserved words or aliases can be removed by quoting any character of the reserved word. But the recognition of function names or built-in command names listed below cannot be altered by quoting them. 👑

Arithmetic Evaluation 🧮

The shell performs arithmetic evaluation for arithmetic expansion, to evaluate an arithmetic command, to evaluate an indexed array subscript, and to evaluate arguments to the built-in commands shift and let, as well as arguments to numeric format specifiers given to print -f and printf. Evaluations use double precision floating point arithmetic — or long double precision floating point on systems that provide that data type. Floating point constants follow the ANSI C programming language floating point conventions. The case-insensitive floating point constants NaN and Inf can be used to represent "not a number" and infinity respectively, unless the posix shell option is on. Integer constants follow the ANSI C programming language integer constant conventions, although only single byte character constants are recognized, and character casts are not recognized.

Also, integer constants can be of the form base#n, or -base#n for negative numbers, where base is a decimal number between two and sixty-four representing the arithmetic base, and n is an unsigned integer in that base. The digits above 9 are represented by the lower case letters, the upper case letters, @, and _, respectively. For bases less than or equal to 36, upper and lower case letters can be used interchangeably. 💅

An arithmetic expression uses the same syntax, precedence, and associativity of expression as the C language. All the C language operators that apply to floating point quantities can be used. On top of that, the operator ** can be used for exponentiation — it has higher precedence than multiplication and it's left associative. And when the value of an arithmetic variable or subexpression can be represented as a long integer, all C language integer arithmetic operations can be performed. Variables can be referenced by name within an arithmetic expression without using the parameter expansion syntax. When a variable is referenced, its value gets evaluated as an arithmetic expression.

Any of the following math library functions that are in the C math library can be used within an arithmetic expression:

abs acos acosh asin asinh atan atan2 atanh cbrt ceil copysign cos cosh
erf erfc exp exp10 exp2 expm1 fabs fdim finite float floor fma fmax fmin
fmod fpclass fpclassify hypot ilogb int isfinite isgreater
isgreaterequal isinf isinfinite isless islessequal islessgreater isnan
isnormal issubnormal isunordered iszero j0 j1 jn ldexp lgamma log log10
log1p log2 logb nearbyint nextafter nexttoward pow remainder rint round
scalb scalbn signbit sin sinh sqrt tan tanh tgamma trunc y0 y1 yn

In addition, arithmetic functions can be defined as shell functions with a variant of the function name syntax,

function .sh.math.name ident ... { list ;}

where name is the function name you'd use inside the arithmetic expression, and each identifier, ident, is a name reference to the long double precision floating point argument. 💖 The value of .sh.value when the function returns — that's the value of this function, bestie. User defined functions can take up to 3 arguments and they override the C math library functions. That's main character energy fr.

So you can also decide a variable is internally a double precision floating point using the -E [n], -F [n], or -X [n] option of the typeset special built-in command. Let's break it down 👇

-E [n] — makes the expansion of the value be represented using scientific notation when it's expanded. The optional option argument n defines the number of significant figures.
-F [n] — makes the expansion be represented as a floating decimal number when it's expanded.
-X [n] — makes the expansion be represented using the %a format defined by ISO C-99. The optional option argument n here defines the number of places after the decimal (or radix) point.

And if you want a variable to be internally an integer, that's the -i [n] option of the typeset special built-in command. The optional option argument n specifies an arithmetic base to use when expanding the variable. Don't specify a base? Base 10, obviously. 💅

Arithmetic evaluation is performed on the value of each assignment to a variable that has the -E, -F, -X, or -i attribute. And heads up 🚨 — assigning a floating point number to a variable whose type is an integer causes the fractional part to be truncated. Gone. Poof. 🫠

✨ Prompting ✨

When you're using it interactively, the shell prompts you with the value of PS1 after expanding it for parameter expansion, command substitution, and arithmetic expansion — and that happens before reading a command. 🌸

Also, each single ! in the prompt gets replaced by the command number. And a !! is required to place ! in the prompt (yes, you have to escape the escape, it's giving syntax drama 😭).

If at any time a new-line is typed and further input is needed to complete a command, then the secondary prompt — i.e., the value of PS2 — is issued. 🎀

🔍 Conditional Expressions 🔍

A conditional expression is used with the [[ compound command to test attributes of files and to compare strings. Field splitting and pathname expansion are not performed on the words between [[ and ]]. Periodt.

Each expression can be constructed from one or more of the following unary or binary expressions:

string — Same as -n string below.
-a file — Same as -e below. This is obsolete.
-b file — True if file exists and is a block special file.
-c file — True if file exists and is a character special file.
-d file — True if file exists and is a directory.
-e file — True if file exists.
-f file — True if file exists and is an ordinary file.
-g file — True if file exists and it has its setgid bit set.
-k file — True if file exists and it has its sticky bit set.
-n string — True if length of string is non-zero.
-o ?option — True if option named option is a valid option name.
-o option — True if option named option is on.
-p file — True if file exists and is a fifo special file or a pipe.
-r file — True if file exists and is readable by current process.
-s file — True if file exists and has size greater than zero.
-t fildes — True if file descriptor number fildes is open and associated with a terminal device.
-u file — True if file exists and it has its setuid bit set.
-v name — True if variable name is a valid variable name and is set.
-w file — True if file exists and is writable by current process.
-x file — True if file exists and is executable by current process. If file exists and is a directory, then true if the current process has permission to search in the directory.
-z string — True if length of string is zero.
-L file — True if file exists and is a symbolic link.
-h file — True if file exists and is a symbolic link.
-N file — True if file exists and the modification time is greater than the last access time.
-O file — True if file exists and is owned by the effective user ID of this process.
-G file — True if file exists and its group matches the effective group ID of this process.
-R name — True if variable name is a name reference.
-S file — True if file exists and is a socket.
file1 -nt file2 — True if file1 exists and file2 does not, or file1 is newer than file2.
file1 -ot file2 — True if file2 exists and file1 does not, or file1 is older than file2.
file1 -ef file2 — True if file1 and file2 exist and refer to the same file.
string == pattern — True if string matches pattern. Any part of pattern can be quoted to cause it to be matched as a string. With a successful match to a pattern, the .sh.match array variable will contain the match and subpattern matches. 👑
string = pattern — Same as == above, but is obsolete.
string != pattern — True if string does not match pattern. When the string matches the pattern the .sh.match array variable will contain the match and subpattern matches.
string =~ ere — True if string matches the pattern ~(E)ere where ere is an extended regular expression.
string1 < string2 — True if string1 comes before string2 based on ASCII value of their characters.
string1 > string2 — True if string1 comes after string2 based on ASCII value of their characters.

The following obsolete arithmetic comparisons are also permitted (they're still valid, just… vintage 💀):

exp1 -eq exp2 — True if exp1 is equal to exp2.
exp1 -ne exp2 — True if exp1 is not equal to exp2.
exp1 -lt exp2 — True if exp1 is less than exp2.
exp1 -gt exp2 — True if exp1 is greater than exp2.
exp1 -le exp2 — True if exp1 is less than or equal to exp2.
exp1 -ge exp2 — True if exp1 is greater than or equal to exp2.

In each of the above expressions, if file is of the form /dev/fd/n, where n is an integer, then the test is applied to the open file whose descriptor number is n. 🔥 The posix shell option disables this special handling.

A compound expression can be constructed from these primitives by using any of the following, listed in decreasing order of precedence:

(expression) — True if expression is true. Used to group expressions.
! expression — True if expression is false.
expression1 && expression2 — True if expression1 and expression2 are both true.
expression1 || expression2 — True if either expression1 or expression2 is true.

💾 Input/Output 💾

Before a command is executed, its input and output may be redirected using a special notation interpreted by the shell. The following may appear anywhere in a simple-command or may precede or follow a command, and they are not passed on to the invoked command.

Command substitution, parameter expansion, and arithmetic expansion occur before word or digit is used — except as noted below. Pathname expansion occurs only if the shell is interactive and the pattern matches a single file. Field splitting is not performed. 🦋

In each of the following redirections, if file is of the form /dev/sctp/host/port, /dev/tcp/host/port, or /dev/udp/host/port, where host is a hostname or host address, and port is a service given by name or an integer port number, then the redirection attempts to make a tcp, sctp or udp connection to the corresponding socket. Networking, in my shell? Iconic. 🌈

No intervening space is allowed between the characters of redirection operators. Like, none. Zero. Not even a lil one. 🚫

<word — Use file word as standard input (file descriptor 0).
>word — Use file word as standard output (file descriptor 1). If the file does not exist then it is created. If the file exists, and the noclobber option is on, this causes an error; otherwise, it is truncated to zero length.
>|word — Same as >, except that it overrides the noclobber option. Rebellious. 💅
>;word — Write output to a temporary file. If the command completes successfully rename it to word, otherwise, delete the temporary file. >;word cannot be used with the exec and redirect built-ins.
>>word — Use file word as standard output. If the file exists, then output is appended to it (by first seeking to the end-of-file); otherwise, the file is created.
<>word — Open file word for reading and writing as standard output. If the posix option is active, it defaults to standard input instead.
<>;word — The same as <>word except that if the command completes successfully, word is truncated to the offset at command completion. <>;word cannot be used with the exec and redirect built-ins.

<<[-]word — okay so this one reads the shell input up to a line that matches word once all quoting has been stripped off it, or until end-of-file, whichever hits first. Ngl, nothing happens to word itself — no parameter expansion, no command substitution, no arithmetic expansion, no pathname expansion, periodt. The resulting document is called a here-document, and it becomes standard input. Here's the tea: if any character of word is quoted, then the shell puts zero interpretation on the characters of the document. But if it's not quoted, then parameter expansion, command substitution and arithmetic expansion all happen, \new-line gets ignored, and you gotta use \ to quote the characters \, $, and `. Add - onto << and every leading tab gets stripped from word and from the document. Add # onto << and leading spaces and tabs get stripped off the first line of the document, and up to an equivalent indentation gets stripped from the remaining lines and from word too — a tab stop is assumed at every 8 columns for figuring out that indentation. 💅
<<<word — the short lil' version of a here document, where word becomes the contents of the here-document after parameter expansion, command substitution and arithmetic expansion all do their thing. ✨
<&digit — standard input gets duplicated from file descriptor digit (peep dup(2)).
>&digit — standard output gets duplicated from file descriptor digit.
<&digit- — the file descriptor given by digit gets moved to standard input.
>&digit- — the file descriptor given by digit gets moved to standard output.
<&- — standard input is closed. bye bestie 👋
>&- — standard output is closed. also bye 👋
<&p — the input from the co-process gets moved to standard input.
>&p — the output to the co-process gets moved to standard output.
<#((expr)) — evaluates the arithmetic expression expr and positions file descriptor 0 to the resulting value, in bytes, from the start of the file. When expr is being evaluated, the variables CUR and EOF evaluate to the current offset and the end-of-file offset respectively. 🎯
>#((offset)) — same as <#, except it applies to file descriptor 1.
<#pattern — seeks forward to the beginning of the next line containing pattern.
<##pattern — same as <#, except the portion of the file that gets skipped is also copied to standard output.

If one of the above is preceded by a digit, with no intervening space, then the file descriptor number it refers to is the one that digit specifies — instead of the default 0 or 1. And if one of the above — other than >&- and the ># and <# forms — is preceded by {varname} with no intervening space, then a file descriptor number > 9 will be picked out by the shell and stored in the variable varname, so you can read from or write to it with redirections like <& $varname or >& $varname. If >&- or any of the ># and <# forms is preceded by {varname}, then the value of varname defines the file descriptor to close or position. Like, for example:

... 2>&1

means file descriptor 2 is to be opened for writing as a duplicate of file descriptor 1, and:

exec {n}<file

means open the file named file for reading and store the file descriptor number in the variable n. iconic behavior fr fr 🫶

A special shorthand redirection operator &>word is available; it's equivalent to >word 2>&1. It cannot be preceded by any digit or variable name. And heads up — this shorthand is disabled if the posix shell option is active. 🙈

The order in which redirections are specified is significant. The shell evaluates each redirection in terms of the (file descriptor, file) association at the time of evaluation. Example time:

... 1>fname 2>&1

first associates file descriptor 1 with the file fname. Then it associates file descriptor 2 with the file associated with file descriptor 1 (i.e. fname). If the order of redirections were reversed, file descriptor 2 would be associated with the terminal (assuming file descriptor 1 had been) and then file descriptor 1 would be associated with the file fname. so basically order is the moment, don't play with it 😭

If a command is followed by & and job control is not active, then the default standard input for the command is the empty file /dev/null. Otherwise, the environment for the execution of a command contains the file descriptors of the invoking shell as modified by input/output specifications.

🌸 Environment 🌸

The environment (see environ(7)) is a list of name-value pairs that gets passed to an executed program the same way a normal argument list would be. The names have to be identifiers and the values are character strings. The shell interacts with the environment in several ways: on invocation, the shell scans the environment and creates a variable for each name it finds, giving it the corresponding value and attributes and marking it export. Executed commands inherit the environment. If the user modifies the values of these variables or creates new ones using the export or typeset -x commands, they become part of the environment. So the environment seen by any executed command is composed of any name-value pairs originally inherited by the shell, whose values may be modified by the current shell, plus any additions which must be noted in export or typeset -x commands. 💖

Exported variables pass their attributes through the environment so that a newly invoked ksh that is a child or exec'd process of the current shell will automatically import them — unless the posix shell option is on. As of ksh 93u+m/1.0, this never includes the readonly attribute.

Heads up bestie: passing attributes through the environment is deprecated and this feature will be removed in ksh 93u+m/1.1. 😭

The environment for any simple-command or function may be augmented by prefixing it with one or more variable assignments. A variable assignment argument is a word of the form identifier=value. So:

TERM=450 cmd args                  and
(export TERM; TERM=450; cmd args)

are equivalent (as far as the above execution of cmd is concerned, except for special built-in commands listed below — those that are marked with †).

If the obsolete -k option is set, all variable assignment arguments get placed in the environment, even if they occur after the command name. The following first prints a=b c and then c:

echo a=b c
set -k
echo a=b c

This feature is intended for use with scripts written for early versions of the shell, and its use in new scripts is strongly discouraged. It is likely to disappear someday. 💀

🧚 Functions 🧚

For historical reasons, there are two ways to define functions: the name() syntax and the function name syntax, described in the Commands section above. Shell functions are read in and stored internally. Alias names are resolved when the function is read. Functions are executed like commands, with the arguments passed as positional parameters. (See Execution below.) 👑

Functions defined by the function name syntax and called by name execute in the same process as the caller and share all files and present working directory with the caller. All changes to shell options are local to the function call. The errexit option is turned off in the function's scope. The same applies to the xtrace option, unless the functrace shell option is set or the function was tagged using typeset -f -t. The state of all other shell options is inherited from the caller. Traps caught by the caller are reset to their default action inside the function. A trap condition that is not caught or ignored by the function causes the function to terminate and the condition to be passed on to the caller. A trap on EXIT set inside a function is executed in the environment of the caller after the function completes. Ordinarily, variables are shared between the calling program and the function. However, the typeset special built-in command used within a function defines local variables whose scope includes the current function. They can be passed to functions that they call in the variable assignment list that precedes the call, or as arguments passed as name references. Errors within functions return control to the caller. ✨

Functions defined with the name() syntax, and functions defined with the function name syntax that are invoked with the . special built-in, are executed in the caller's environment and share all variables, options and traps with the caller. Errors within these function executions cause the script that contains them to abort. 🚨

The special built-in command return is what you use to return from function calls.

Function names can be listed with the -f or +f option of the typeset special built-in command. The text of functions, when available, will also be listed with -f. Functions can be undefined with the -f option of the unset special built-in command.

Ordinarily, functions are unset when the shell executes a shell script. Functions that need to be defined across separate invocations of the shell should be placed in a directory, and the FPATH variable should contain the name of this directory. They may also be specified in the ENV file. 🦋

💜 Discipline Functions 💜

Each variable can have zero or more discipline functions associated with it. The shell initially understands the discipline names get, set, append, and unset, but more can be added when defining new types. On most systems others can be added at run time via the C programming interface extension provided by the builtin built-in utility. If the get discipline is defined for a variable, it is invoked whenever the given variable is referenced. If the variable .sh.value is assigned a value inside the discipline function, the referenced variable will evaluate to this value instead. If the set discipline is defined for a variable, it is invoked whenever the given variable is assigned a value. If the append discipline is defined for a variable, it is invoked whenever a value is appended to the given variable. The variable .sh.value is given the value of the variable before invoking the discipline, and the variable will be assigned the value of .sh.value after the discipline completes. If .sh.value is unset inside the discipline, then that value is unchanged. If the unset discipline is defined for a variable, it is invoked whenever the given variable is unset. 🌈

okay so the variable .sh.name holds the name of the variable whose discipline function just got called ✨, .sh.subscript is that variable's subscript, and .sh.value is gonna hold the value being assigned inside the set discipline function. The variable _ is a reference to the variable itself, subscript and all if there is one 💅. And for the set discipline specifically — changing .sh.value changes the value that actually gets assigned. no cap. Finally, the expansion ${var.name}, when name is the name of a discipline and there's no variable with that name, is equivalent to the command substitution ${ var.name;}. periodt.

🎀 Name Spaces 🎀

So commands and functions that get executed as part of the list of a namespace command and that modify variables or create new ones — they create a new variable whose name is the name of the name space as given by identifier preceded by a . 💫. When a variable whose name is name gets referenced, it's searched for first using .identifier.name. Same energy for functions: a function defined by a command in the namespace list is created using the name space name preceded by a ..

When the list of a namespace command contains another namespace command, the names of the variables and functions that get created consist of the variable or function name preceded by the list of identifiers, each one preceded by a . 🦋.

Outside of a name space, a variable or function created inside a name space can be referenced by preceding it with the name space name. iykyk.

By default, variables starting with .sh are in the sh name space. 🫶

🌸 Type Variables 🌸

Typed variables are the way to create data structures and objects, bestie. A type can be defined by a shared library, by the enum built-in command described below, or by using the new -T option of the typeset built-in command. With the -T option of typeset, the type name — specified as an option argument to -T — is set with a compound variable assignment that defines the type. Function definitions can show up inside the compound variable assignment, and these become discipline functions for this type, invokable or redefinable by each instance of the type. The function name create is treated specially — it's invoked for each instance of the type that gets created, but it is not inherited and cannot be redefined for each instance. 👑

When a type is defined, a special built-in command of that name gets added. These built-ins are declaration commands and follow the same expansion rules as the built-in commands described below that are marked with a ‡ symbol. These commands can subsequently be used inside further type definitions. The man page for these commands can be generated by using the --man option or any of the other -- options described with getopts. The -r, -a, -A, -h, and -S options of typeset are permitted with each of these new built-ins. ✨

An instance of a type is created by invoking the type name followed by one or more instance names. Each instance of the type is initialized with a copy of the subvariables — except for subvariables defined with the -S option. Variables defined with -S are shared by all instances of the type. Each instance can change the value of any subvariable and can also define new discipline functions with the same names as those defined by the type definition, plus any standard discipline names. But ngl — no additional subvariables can be defined for any instance. 💀

When you're defining a type, if the value of a subvariable is not set and the -r attribute is specified, it makes the subvariable a required subvariable. Whenever an instance of a type is created, all required subvariables must be specified. These subvariables become read-only in each instance. 🎀

When unset is invoked on a subvariable within a type, and the -r attribute has not been specified for this field, the value is reset to the default value associated with the type. Invoking unset on a type instance that is not contained within another type deletes all subvariables and the variable itself. gone. poof. 🦋

A type definition can be derived from another type definition by defining the first subvariable name as _ and defining its type as the base type. Any remaining definitions are additions and modifications that apply to the new type. If the new type name is the same as the base type's name, the type will be replaced and the original type will no longer be accessible 😭.

The typeset command with the -T and no option argument or operands will write all the type definitions to standard output in a form that can be read in to create all the types. slay. 💅

💅 Jobs 💅

If the monitor option of the set command is turned on, an interactive shell associates a job with each pipeline. It keeps a table of current jobs, printed by the jobs command, and assigns them small integer numbers. When a job is started asynchronously with &, the shell prints a line that looks like:

[1] 1234

indicating that the job which was started asynchronously was job number 1 and had one (top-level) process, whose process ID was 1234. ✨

If you're running a job and you wanna do something else, you may hit the key ^Z (control-Z) which sends a STOP signal to the current job. The shell will then normally indicate that the job has been Stopped, and print another prompt. You can then manipulate the state of this job — putting it in the background with the bg command, or running some other commands and then eventually bringing the job back into the foreground with the foreground command fg. A ^Z takes effect immediately and is like an interrupt in that pending output and unread input are discarded when it is typed. 🫶

A job being run in the background will stop if it tries to read from the terminal. Background jobs are normally allowed to produce output, but this can be disabled by giving the command stty tostop. If you set this tty option, then background jobs will stop when they try to produce output, like they do when they try to read input. 😭

There are several ways to refer to jobs in the shell. A job can be referred to by the process ID of any process of the job, or by one of the following:

%number — the job with the given number.
%string — any job whose command line begins with string.
%?string — any job whose command line contains string.
%% — current job.
%+ — equivalent to %%.
%- — previous job.

The shell learns immediately whenever a process changes state. It normally informs you whenever a job becomes blocked so that no further progress is possible, but only just before it prints a prompt. This is done so that it doesn't otherwise disturb your work. The notify option of the set command causes the shell to print these job change messages as soon as they occur. If this happens while you're typing a command with one of the built-in line editors active, the job change message appears above your input without disturbing your command entry. iconic. ✨

When the monitor option is on, each background job that completes triggers any trap set for CHLD. 💫

When you try to leave the shell while jobs are running or stopped, you will be warned that You have stopped(running) jobs. You may use the jobs command to see what they are. If you immediately try to exit again, the shell will not warn you a second time, and the stopped jobs will be terminated 💀. When a login shell receives a HUP signal, it sends a HUP signal to each job that has not been disowned with the disown built-in command described below.

⭐ Signals ⭐

The INT and QUIT signals for an invoked command are ignored if the command is followed by & and the monitor option is not active. Otherwise, signals have the values inherited by the shell from its parent (but see also the trap built-in command below). 🎀

🔥 Execution 🔥

Each time a command is read, the above expansions and substitutions are carried out. If the command name matches one of the Special Built-in Commands listed below, it is executed within the current shell process. Next, the command name is checked to see if it matches a user defined function. If it does, the positional parameters are saved and then reset to the arguments of the function call. A function is also executed in the current shell process. When the function completes or issues a return, the positional parameter list is restored. For functions defined with the function name syntax, any trap set on EXIT within the function is executed. The exit value of a function is the value of the last command executed. If a command name is not a special built-in command or a user defined function, but it is one of the built-in commands listed below, it is executed in the current shell process. periodt. 💅

The shell variables PATH followed by the variable FPATH define the list of directories to search for the command name. Alternative directory names are separated by a colon (:). The default path is the value that was output by getconf PATH at the time ksh was compiled. The current directory can be specified by two or more adjacent colons, or by a colon at the beginning or end of the path list. If the command name contains a /, then the search path is not used. Otherwise, each directory in the list of directories defined by PATH and FPATH is checked in order. If the directory being searched is contained in FPATH and contains a file whose name matches the command being searched, then this file is loaded into the current shell environment as if it were the argument to the . command — except that only preset aliases are expanded, and a function of the given name is executed as described above. 🦋

If this directory is not in FPATH, the shell first determines whether there's a built-in version of a command corresponding to a given pathname, and if so it is invoked in the current process. If no built-in is found, the shell checks for a file named .paths in this directory. If found, and there's a line of the form FPATH=path where path names an existing directory, then that directory is searched immediately after the current directory as if it were found in the FPATH variable. If path does not begin with /, it is checked for relative to the directory being searched. 🌸

The .paths file is then checked for a line of the form PLUGIN_LIB=libname [ : libname ] .... Each library named by libname will be searched for as if it were an option argument to builtin -f, and if it contains a built-in of the specified name, this will be executed instead of a command by this name. Any built-in loaded from a library found this way will be associated with the directory containing the .paths file, so it will only execute if not found in an earlier directory. 👑

Finally, the directory gets checked for a file of the given name. If that file has execute permission but is not an a.out file, it's assumed to be a file containing shell commands, and a whole separate shell gets spawned just to read it. 💫 In that case, all non-exported variables are removed. If the .paths file contains a line of the form name=value in the first or second line, then the environment variable name gets modified by prepending the directory specified by value to the directory list. And if value is not an absolute directory, then it specifies a directory relative to the directory that the executable was found in. If the environment variable name doesn't already exist, it will be added to the environment list for the specified command. A parenthesized command is executed in a subshell without removing non-exported variables. 🫶

🎀 Command Re-entry 🎀

The text of the last HISTSIZE (default 512) commands entered from a terminal device is saved in a history file. The file $HOME/.sh_history is used if the HISTFILE variable is not set, or if the file it names is not writable. A shell can access the commands of all interactive shells which use the same named HISTFILE. 👑

The built-in command hist is used to list or edit a portion of this file. The portion of the file to be edited or listed can be selected by number, or by giving the first character or characters of the command. A single command or a range of commands can be specified.

If you don't specify an editor program as an argument to hist, then the value of the variable HISTEDIT is used. If HISTEDIT is unset, the obsolete variable FCEDIT is used. If FCEDIT is not defined, then /bin/ed is used. The edited command(s) is printed and re-executed upon leaving the editor — unless you quit without writing, obviously. 🙈

The -s option (and in obsolete versions, the editor name -) is used to skip the editing phase and re-execute the command. In this case a substitution parameter of the form old=new can be used to modify the command before execution. For example, with the preset alias r, which is aliased to hist -s, typing r bad=good c will re-execute the most recent command which starts with the letter c, replacing the first occurrence of the string bad with the string good. chef's kiss, no cap ✨

🦋 History Expansion 🦋

History expansions introduce words from the history list into the input stream, which makes it so easy to repeat commands, repeat arguments of a previous command in the current command, or fix typos in the previous command. The history expansion facility is an alternative to history control via the fc or hist built-in command. To enable it, turn on the -H or histexpand option using the set command (see Built-in Commands below). 💅

History expansions begin with the character !. They may begin anywhere in the input. The ! is passed unchanged, with its special meaning disabled, when preceded by \ or ${, enclosed in single quotes, or followed by a space, tab, newline, = or (. History expansions do not nest. They are parsed separately before the shell parser is invoked, so they can override shell grammar rules. 🤯

By default, the expanded version of any line that contains a history expansion is printed, added to the history, and then immediately executed. History expansions are never added to the history themselves, regardless of whether they succeed or fail due to an error. Normally this means that a command line with an erroneous history expansion is just lost and needs to be retyped from scratch 😭 — but if the histreedit shell option is turned on and a line editor is active (see In-line Editing Options below), the erroneous line is pre-filled into the next prompt's input buffer for correcting. The histverify option causes the same thing to be done for the results of successful history expansions, letting you verify and edit before execution. 🫶

A history expansion may have an event specification, which indicates the event from which words are to be taken, a word designator, which selects particular words from the chosen event, and/or a modifier, which manipulates the selected words.

An event specification can be:

n — a number, referring to a particular event.
-n — an offset, referring to the event n before the current event.
# — the current event.
! — the previous event (equivalent to -1).
s — the most recent event whose first word begins with the string s.
?s? — the most recent event which contains the string s. The second ? can be omitted if it is immediately followed by a newline.

For example, consider this bit of someone's history list, as might be output by the hist -l command:

9        nroff -man wumpus.man
10       cp wumpus.man wumpus.man.old
11       vi wumpus.man
12       diff wumpus.man.old wumpus.man

The commands are shown with their event numbers. The current event, which we haven't typed in yet, is event 13. !11 and !-2 refer to event 11. !! refers to the previous event, 12. !! can be abbreviated ! if it is followed by : (see below). !n refers to event 9, which begins with n. !?old? also refers to event 12, which contains old. Without word designators or modifiers, history references simply expand to the entire event, so we might type !cp to redo the copy command, or !!|more if the diff output scrolled off the top of the screen. 🌈

To select words from an event, the event specification can be followed by a : and a designator for the desired words. The words of an input line are numbered from 0 — the first word (usually the command name) being 0, the second word (first argument) being 1, etc. The basic word designators are:

0 — the first word (command name).
n — the nth argument.
^ — the first argument, equivalent to 1.
$ — the last argument.
% — the word matched by the most recent ?s? search.
x-y — a range of words.
-y — equivalent to 0-y.
* — equivalent to ^-$, but returns nothing if the event contains only 1 word.
x* — equivalent to x-$.
x- — equivalent to x*, but omitting the last word ($).

Selected words are inserted into the command line separated by single blanks. For example, the diff command in the previous example might have been typed as diff !!:1.old !!:1 (using :1 to select the first argument from the previous event) or diff !-2:2 !-2:1 to select and swap the arguments from the cp command. If we didn't care about the order of the diff, we might have said diff !-2:1-2 or simply diff !-2:*. The cp command might have been written cp wumpus.man !#:1.old, using # to refer to the current event. !n:- hurkle.man would reuse the first two words from the nroff command to say nroff -man hurkle.man. 🧚

The : separating the event specification from the word designator can be omitted if the argument selector begins with a ^, $, *, % or -. For example, our diff command might have been diff !!^.old !!^ or, equivalently, diff !!$.old !!$. However, if !! is abbreviated !, an argument selector beginning with - will be interpreted as an event specification. ⚠️

The word(s) in a history reference can be edited by following them with one or more modifiers, each preceded by a colon (:):

h — remove a trailing pathname component, leaving the head.
t — remove all leading pathname components, leaving the tail.
r — remove a filename extension .xxx, leaving the root name.
e — remove all but the extension.
s/l/r/ — substitute l for r. l is simply a string like r, not a regular expression as in the eponymous ed(1) command. Any character may be used as the delimiter in place of /; a \ can be used to quote the delimiter inside l and r. The character & in the r is replaced by l; \ also quotes &. If l is empty, the l from the previous substitution is used, or if there is none, the s from the most recent ?s? search. The trailing delimiter may be omitted if it is immediately followed by a newline.
& — repeat the previous substitution.
g — global substitution, for example :gs/foo/bar/ or :g&. Applies the s or & modifier to the entire command line.
a — same as g.
p — print the new command line but do not execute it.
q — quote the expanded words, preventing further expansions.
x — like q, but break into words at blanks, tabs and newlines.

Modifiers are applied to only the first modifiable word (unless g or a is used). It is an error for no word to be modifiable. 💀

For example, the diff command might have been written as diff wumpus.man.old !#^:r, using :r to remove .old from the first argument on the same line (!#^). We might follow mail -s "I forgot my password" rot with !:s/rot/root to correct the spelling of root. ✨

History expansions also occur when an input line begins with ^. When it is the first character on an input line, it is an abbreviation of !:s^. Thus we might have said ^rot^root to make the spelling correction in the previous example. This is the only history expansion that does not explicitly begin with !. 🎀

If a word on a command line begins with the history comment character #, history expansion is ignored for the rest of that line. This usually causes the shell parser (which uses the same character to signal a comment) to treat the rest of the line as a comment as well — but since history expansion is parsed separately from the shell grammar and with different rules, this cannot be guaranteed in all cases. And if the history comment character is changed, the shell grammar comment character does not change along with it. 🫠

The three characters used to signal history expansion can be changed using the histchars shell variable; see Shell Variables above.

💅 In-line Editing Options 💅

Normally, each command line entered from a terminal device is simply typed followed by a new-line (RETURN or LINE FEED). But if either the emacs, gmacs, or vi option is active, the user can edit the command line. To be in any of these edit modes, set the corresponding option. An editing option is automatically selected each time the VISUAL or EDITOR variable is assigned a value matching any of these editor names; for details, see Shell Variables above under VISUAL. 👑

The editing features require that the user's terminal accept RETURN as carriage return without line feed, and that a space ( ) must overwrite the current character on the screen. important vibe check, bestie ⭐

Unless the multiline option is on, the editing modes are giving "looking through a window at the current line" 💅. The window width is the value of COLUMNS if it's defined, otherwise 80. If the window width is too small to display the prompt and leave at least 8 columns to enter input, the prompt gets truncated from the left 😭. If the line is longer than the window width minus two, a mark shows up at the end of the window to let you know. As the cursor moves and hits the window boundaries, the window gets centered about the cursor. The mark is a > (<, *) if the line extends on the right (left, both) side(s) of the window ✨.

The search commands in each edit mode give you access to the history file 🦋. Only strings are matched, not patterns, although a leading ^ in the string restricts the match to begin at the first character in the line.

Each of the edit modes has an operation to list the files or commands that match a partially entered word 👑. When it's applied to the first word on the line, or the first word after a ;, |, &, or (, and the word does not begin with ~ or contain a /, the list of aliases, functions, and executable commands defined by the PATH variable that could match the partial word is displayed 💖. Otherwise, the list of files that match the given word is displayed. If the partially entered word does not contain any file expansion characters, a * is appended before generating these lists. After displaying the generated list, the input line is redrawn. These operations are called command name listing and file name listing, respectively.

There are additional operations, referred to as command name completion and file name completion, which compute the list of matching commands or files, but instead of printing the list, replace the current word with a complete or partial match 🎀. For file name completion, if the match is unique, a / is appended if the file is a directory and a space is appended if the file is not a directory. Otherwise, the longest common prefix for all the matching files replaces the word. For command name completion, only the portion of the file names after the last / are used to find the longest command prefix. If only a single name matches this prefix, then the word is replaced with the command name followed by a space ✨. When using a tab for completion that does not yield a unique match, a subsequent tab will provide a numbered list of matching alternatives. A specific selection can be made by entering the selection number followed by a tab. Neither completion nor listing operations are attempted before the first character in a line 🙈.

🔑 Key Bindings 🔑

The KEYBD trap can be used to intercept keys as they are typed and change the characters that are actually seen by the shell 👀. This trap is executed after each character (or sequence of characters when the first character is ESC) is entered while reading from a terminal. The variable .sh.edchar contains the character or character sequence which generated the trap. Changing the value of .sh.edchar in the trap action causes the shell to behave as if the new value were entered from the keyboard rather than the original value 💅.

The variable .sh.edcol is set to the input column number of the cursor at the time of the input. The variable .sh.edmode is set to ESC when in vi input mode (see below) and set to the empty string otherwise. Prepending ${.sh.editmode} to a value assigned to .sh.edchar will cause the shell to change to control mode if it is not already in this mode 🌈.

This trap is not invoked for characters entered as arguments to editing directives, or while reading input for a character search 🚫

💅 Emacs Editing Mode 💅

This mode is entered by enabling either the emacs or gmacs option 👑. The only difference between these two modes is the way they handle ^T. To edit, you move the cursor to the point needing correction and then insert or delete characters or words as needed. All the editing commands are control characters or escape sequences 🎀.

The notation for control characters is caret (^) followed by the character. For example, ^F is the notation for control F. This is entered by depressing f while holding down the CTRL (control) key. The SHIFT key is not depressed. (The notation ^? indicates the DEL (delete) key.) 🌸

The notation for escape sequences is M- followed by a character. For example, M-f (pronounced Meta f) is entered by depressing ESC (ASCII 033) followed by f. (M-F would be the notation for ESC followed by SHIFT (capital) F.) ✨

All edit commands operate from any place on the line (not just at the beginning) 👏. Neither the RETURN nor the LINE FEED key is entered after edit commands except when noted.

The M-[ multi-character commands below are DEC VT220 escape sequences generated by special keys on standard PC keyboards, such as the arrow keys. You could type them directly but they are meant to recognize the keys in question, which are indicated in parentheses 🧚

^F — slide the cursor forward (right) by one character. 💅
M-[C — (Right arrow) literally the same thing as ^F.
M-f — hop the cursor forward one whole word. And bestie, the emacs editor's idea of a word is a string of characters consisting of only letters, digits and underscores. that's it. that's the vibe.
M-[1;3C — (Alt-Right arrow) same as M-f.
M-[1;5C — (Ctrl-Right arrow) same as M-f.
M-[1;9C — (iTerm2 Alt-Right arrow) same as M-f.
^B — slide the cursor backward (left) one character.
M-[D — (Left arrow) same as ^B.
M-b — hop the cursor backward one word.
M-[1;3D — (Alt-Left arrow) same as M-b.
M-[1;5D — (Ctrl-Left arrow) same as M-b.
M-[1;9D — (iTerm2 Alt-Left arrow) same as M-b.
^A — yeet the cursor to the start of the line.
M-[H — (Home) same as ^A.
M-[1~ — same as ^A.
M-[7~ — same as ^A.
^E — yeet the cursor to the end of the line.
M-[F — (End) same as ^E.
M-[4~ — same as ^E.
M-[8~ — same as ^E.
M-[Y — same as ^E.
M-OA — (Up Arrow) same as M-[A.
M-OB — (Down Arrow) same as M-[B.
M-OC — (Right Arrow) same as M-[C.
M-OD — (Left Arrow) same as M-[D.
M-O5C — (Ctrl-Right Arrow) same as M-f.
M-O5D — (Ctrl-Left Arrow) same as M-b.
^]char — move the cursor forward to the character char on the current line.
M-^]char — move the cursor backward to the character char on the current line.
^X^X — swap the cursor and the mark. exchanging positions, iconic behavior honestly 💫
erase — the user defined erase character as defined by the stty(1) command, usually ^H. deletes the previous character.
lnext — the user defined literal next character as defined by the stty(1) command, or ^V if that's not defined. strips the next character's editing features (if any) right off. 🧼
^D — delete the current character.
M-[3~ — (Forward delete) same as ^D.
M-d — delete the current word.
M-[3;5~ — (Ctrl-Delete) same as M-d.
M-^H — (Meta-backspace) delete the previous word.
M-h — delete the previous word.
M-^? — (Meta-DEL) delete the previous word — but heads up: if your interrupt character is ^? (DEL, which is the default) then this command will not work. 😭
^T — transpose the current character with the previous character and advance the cursor, in emacs mode. in gmacs mode it transposes the two previous characters instead. slay, I guess.
^C — capitalize the current character.
M-c — capitalize the current word.
M-l — change the current word to lower case.
^K — delete from the cursor to the end of the line. if it's preceded by a numerical parameter whose value is less than the current cursor position, then it deletes from that given position up to the cursor instead. and if it's preceded by a numerical parameter whose value is greater than the current cursor position, then it deletes from the cursor up to that given cursor position. choose your fighter 🫶
^W — kill from the cursor to the mark.
M-p — push the region from the cursor to the mark onto the stack.
kill — the user defined kill character as defined by the stty(1) command, usually ^U. kills the entire current line. if two kill characters are entered in succession, then from then on all kill characters cause a line feed — which is genuinely useful when you're on a paper terminal, no cap. and a subsequent pair of kill characters undoes that change. ✨
^Y — restore the last item that was removed from the line. (yank item back to the line, as the ancestors intended.)
^X^E — return the command hist -e ${VISUAL:-${EDITOR:-vi}} in the input buffer, to call a full editor — vi by default — on the current command line. 💖
^L — line feed and print the current line.
M-^L — clear the screen. ✨ clean slate energy.
^@ — (Null character) set mark.
M-space — (Meta space) set mark.
^J — (New line) execute the current line.
^M — (Return) execute the current line.
eof — the end-of-file character, normally ^D, is processed as an end-of-file only if the current line is empty. the conditions, bestie. the conditions. 🙈
^P — fetch the previous command. every single time you hit ^P you access the previous command further back in time. it also moves back one line when you're not on the first line of a multi-line command.
M-[A — (Up arrow) if the cursor is at the end of the line, this is equivalent to ^R with string set to the contents of the current line. otherwise, it's equivalent to ^P.
M-< — fetch the least recent (oldest) history line.
M-> — fetch the most recent (youngest) history line.
^N — fetch the next command line. every time you hit ^N you access the next command line forward in time.
M-[B — (Down arrow) equivalent to ^N.
^Rstring — reverse search history for a previous command line containing string. if a parameter of zero is given, the search goes forward instead. string is terminated by a RETURN or NEW LINE. if string is preceded by a ^, the matched line must begin with string. if string is omitted, then the next command line containing the most recent string is accessed — and in that case, a parameter of zero reverses the direction of the search. the lore is deep but we love her 🔍
^G — exit reverse search mode. (the escape hatch, ty)
^O — Operate — execute the current line and fetch the next line relative to the current line from the history file.
M-digits — (Escape) define a numeric parameter; the digits are taken as a parameter to the next command. the commands that accept a parameter are ^F, ^B, erase, ^C, ^D, ^K, ^R, ^P, ^N, ^], M-., M-^], M-_, M-=, M-b, M-c, M-d, M-f, M-h, M-l, M-^H, and the arrow keys and forward-delete key. 😤 memorize it, it's giving syllabus
M-letter — Soft-key — your alias list is searched for an alias by the name _letter and if an alias of this name is defined, its value gets inserted on the input queue. the letter must not be one of the above meta-functions.
M-[letter — Soft-key — your alias list is searched for an alias by the name __letter and if an alias of this name is defined, its value gets inserted on the input queue. this is how you program function keys on many terminals. 🎹
M-. — the last word of the previous command is inserted on the line. if preceded by a numeric parameter, the value of that parameter determines which word to insert instead of the last word.
M-_ — same as M-..
M-* — attempt pathname expansion on the current word. an asterisk gets appended if the word doesn't match any file or doesn't contain any special pattern characters.
M-ESC — command or file name completion as described above.
^I tab — attempts command or file name completion as described above. if a partial completion occurs, repeating this behaves as if M-= were entered. if no match is found, or it's entered after a space, a tab is inserted.
M-= — if not preceded by a numeric parameter, it generates the list of matching commands or file names as described above. otherwise, the word under the cursor is replaced by the item corresponding to the value of the numeric parameter from the most recently generated command or file list — and if the cursor is not on a word, it's inserted instead.
^U — multiply the parameter of the next command by 4.
\ — if the backslashctrl shell option is on (which is the default setting), this escapes the next character. editing characters, and the user's erase, kill and interrupt (normally ^C) characters may be entered in a command line or in a search string if preceded by a \. the \ removes the next character's editing features (if any). see also lnext, which is not subject to any shell option. 🫶
M-^V — display the version of the shell.
M-[2~ — (Insert) escape the next character.
M-# — if the line does not begin with a #, a # is inserted at the beginning of the line and after each new-line, and the line is entered. this causes a comment to be inserted in the history file. if the line begins with a #, the # is deleted and one # after each new-line is also deleted. 💅

🦋 Vi Editing Mode 🦋

okay so there are two typing modes here. initially, when you enter a command, you're in the input mode. to edit, the user enters control mode by typing ESC (033), moves the cursor to the point that needs correcting, and then inserts or deletes characters or words as needed. most control commands accept an optional repeat count prior to the command. very demure, very mindful.

the notation for control characters used below is ^ followed by a character. for instance, ^H is entered by holding down the Control key and pressing H. ^[ (Control+[) is equivalent to the ESC key. the notation for escape sequences is ^[ followed by one or more characters.

the ^[[ (ESC [) multi-character commands below are DEC VT220 escape sequences generated by special keys on standard PC keyboards, such as the arrow keys, which are indicated in parentheses. when in input mode, these keys will switch you to control mode before performing the associated action. these sequences can use preceding repeat count parameters, but only when the ^[ and the subsequent [ are entered into the input buffer at the same time, such as when pressing one of those keys. ⚠️ important detail, don't skip it.

🌸 Input Edit Commands 🌸

by default the editor is in input mode.

erase — the user defined erase character as defined by the stty(1) command, usually ^H or #. deletes the previous character.
^W — delete the previous blank separated word.
eof — as the first character of the line, this causes the shell to terminate unless the ignoreeof option is set. otherwise this character is ignored.
lnext — the user defined literal next character as defined by the stty(1), or ^V if that's not defined. removes the next character's editing features (if any).
\ — if the backslashctrl shell option is on (which is the default setting), this escapes the next erase or kill character.
^I tab — attempts command or file name completion as described above and returns to input mode. if a partial completion occurs, repeating this behaves as if = were entered from control mode. if no match is found, or it's entered after a space, a tab is inserted.

🫶 Motion Edit Commands 🫶

these commands will move the cursor.

[count]l — cursor forward (right) one character. slay ✨
[count]^[[C — (Right arrow) same as l. no cap
[count]w — cursor forward one alphanumeric word. periodt
[count]W — cursor to the beginning of the next word that follows a blank.
[count]e — cursor to end of word.
[count]E — cursor to end of the current blank delimited word.
[count]h — cursor backward (left) one character.
[count]^[[D — (Left arrow) same as h.
[count]b — cursor backward one word.
[count]B — cursor to preceding blank separated word.
[count]| — cursor to column count.
[count]fc — find the next character c in the current line.
[count]Fc — find the previous character c in the current line.
[count]tc — equivalent to f followed by h.
[count]Tc — equivalent to F followed by l.
[count]; — repeats count times, the last single character find command, f, F, t, or T.
[count], — reverses the last single character find command count times.
0 — cursor to start of line.
^[[H — (Home) same as 0.
^[[1~ — same as 0.
^[[7~ — same as 0.
^[[1;3D — (Alt-Left arrow) same as b.
^[[1;5D — (Ctrl-Left arrow) same as b.
^[[1;9D — (iTerm2 Alt-Left arrow) same as b.
^[[1;3C — (Alt-Right arrow) same as w.
^[[1;5C — (Ctrl-Right arrow) same as w.
^[[1;9C — (iTerm2 Alt-Right arrow) same as w.
^[[2~ — (Insert) same as i.
^[[3;5~ — (Ctrl-Delete) same as dw.
^[OA — (Up Arrow) same as ^[[A.
^[OB — (Down Arrow) same as ^[[B.
^[OC — (Right Arrow) same as ^[[C.
^[OD — (Left Arrow) same as ^[[D.
^[O5C — (Ctrl-Right Arrow) same as w.
^[O5D — (Ctrl-Left Arrow) same as b.
^ — cursor to first non-blank character in line.
$ — cursor to end of line.
^[[F — (End) same as $.
^[[4~ — same as $.
^[[8~ — same as $.
^[[Y — same as $.
^G — exit reverse search mode.
% — moves to balancing (, ), {, }, [, or ]. If cursor is not on one of the above characters, the remainder of the line is searched for the first occurrence of one of the above characters first. iykyk

🔍 Search Edit Commands

These commands access your command history. bestie, they're your time machine ⏳✨

[count]k — fetch previous command. Each time k is entered the previous command back in time is accessed.
[count]- — equivalent to k.
[count]^[[A — (Up arrow) if cursor is at the end of the line it is equivalent to / with string set to the contents of the current line. Otherwise, it is equivalent to k.
[count]j — fetch next command. Each time j is entered the next command forward in time is accessed.
[count]+ — equivalent to j.
[count]^[[B — (Down arrow) equivalent to j.
[count]G — the command number count is fetched. The default is the least recent history command.
/string — search backward through history for a previous command containing string. string is terminated by a `RETURN' or `NEW LINE'. If string is preceded by a ^, the matched line must begin with string. If string is empty, the previous string will be used.
?string — same as / except that search will be in the forward direction.
n — search for next match of the last pattern to / or ? commands.
N — search for next match of the last pattern to / or ?, but in reverse direction.

✏️ Text Modification Edit Commands

These commands will modify the line. fr fr, be careful — they're not just for looking cute 💅

a — enter input mode and enter text after the current character.
A — append text to the end of the line. Equivalent to $a.
[count]cmotion, c[count]motion — delete current character through the character that motion would move the cursor to and enter input mode. If motion is c, the entire line will be deleted and input mode entered.
C — delete the current character through the end of line and enter input mode. Equivalent to c$.
S — equivalent to cc.
[count]s — replace characters under the cursor in input mode.
D — delete the current character through the end of line. Equivalent to d$.
[count]dmotion, d[count]motion — delete current character through the character that motion would move to. If motion is d , the entire line will be deleted.
i — enter input mode and insert text before the current character.
I — insert text before the first non-blank character. Equivalent to ^i.
[count]P — place the previous text modification before the cursor.
[count]p — place the previous text modification after the cursor.
R — enter input mode and replace characters on the screen with characters you type overlay fashion.
[count]rc — replace the count character(s) starting at the current cursor position with c, and advance the cursor.
[count]x — delete current character.
[count]^[[3~ — (Forward delete) same as x.
[count]X — delete preceding character.
[count]. — repeat the previous text modification command.
[count]∼ — invert the case of the count character(s) starting at the current cursor position and advance the cursor.
[count]_ — causes the count word of the previous command to be appended and input mode entered. The last word is used if count is omitted.
* — causes an * to be appended to the current word and pathname expansion attempted. If no match is found, it rings the bell. Otherwise, the word is replaced by the matching pattern and input mode is entered.
\ — command or file name completion as described above.

🎀 Other Edit Commands

Miscellaneous commands. a lil grab-bag, bestie 🍓

[count]ymotion, y[count]motion — yank current character through character that motion would move the cursor to and puts them into the delete buffer. The text and cursor are unchanged.
yy — yanks the entire line.
Y — yanks from current position to end of line. Equivalent to y$.
u — undo the last text modifying command.
U — undo all the text modifying commands performed on the line.
[count]v — returns the command hist -e ${VISUAL:-${EDITOR:-vi}} count in the input buffer to call a full editor — vi by default — on a history entry. If count is omitted, then the current line is used.
^L — line feed and print current line. Has effect only in control mode.
^J — (New line) execute the current line, regardless of mode.
^M — (Return) execute the current line, regardless of mode.
# — if the first character of the command is a #, then this command deletes this # and each # that follows a newline. Otherwise, sends the line after inserting a # in front of each line in the command. Useful for causing the current line to be inserted in the history as a comment and uncommenting previously commented commands in the history file.
[count]= — if count is not specified, it generates the list of matching commands or file names as described above. Otherwise, the word under the cursor is replaced by the count item from the most recently generated command or file list. If the cursor is not on a word, it is inserted instead.
@letter — your alias list is searched for an alias by the name _letter and if an alias of this name is defined, its value will be inserted on the input queue for processing.
^V — display version of the shell.

🛠️ Built-in Commands. 🛠️

The simple-commands listed below are built in to the shell and are executed in the same process as the shell. The effects of any added Input/Output redirections are local to the command, except for the exec and redirect commands. Unless otherwise indicated, the output is written on standard output (file descriptor 1) and the exit status, when there is no syntax error, is zero. Except for :, true, false, and echo, all built-in commands accept -- to indicate end of options, and are self-documenting. periodt 💚

The self-documenting commands interpret the option --man as a request to display that command's own manual page, --help as a request to display the OPTIONS section from their manual page, and -? as a request to print a brief usage message. All these are processed as error messages, so they are written on standard error (file descriptor 2) and to pipe them into a pager such as more(1) you need to add a 2>&1 redirection before the |. The display of boldface text depends on whether standard error is on a terminal, so is disabled when using a pager. Exporting the ERROR_OPTIONS environment variable with a value containing emphasis will force this on; a value containing noemphasis forces it off. The test/[ command needs an additional -- argument to recognize self-documentation options, e.g. test --man --. The exec and redirect commands, as they make redirections permanent, should use self-documentation options in a subshell when redirecting, for example: (redirect --man) 2>&1. There are advanced output options as well; see getopts --man for more information. you got this 🦋

Okay so before we even look at the commands, bestie, you need to know about two little symbols that change literally everything, because they're giving main character energy and the shell treats them different. 🫶

Commands that have a † in front of them are special built-in commands, and they get treated specially in these four ways:

Variable assignment lists that come before the command stay in effect when the command finishes. ✨
I/O redirections are processed after the variable assignments.
Errors cause a script that contains them to abort. 💀
They are not valid function names.

And commands with a ‡ in front are declaration commands. Any words that follow them in variable-assignment format get expanded with the exact same rules as a variable assignment. So that means: tilde expansion happens after the =, array assignments of the form varname=(assign_list) are supported, and field splitting and pathname expansion are not performed. iykyk 💅

† : [ arg ... ] — this one only expands parameters. That's it. That's the whole job. Slay, honestly. 🎀
† . name [ arg ... ] — okay so if name is a function defined with the function name reserved word syntax, the function runs in the current environment (as if it had been defined with the name() syntax instead). Otherwise, if name refers to a file, the file gets read in its entirety and the commands run in the current shell environment. The search path specified by PATH is used to find the directory containing the file. If any arguments arg are given, they become the positional parameters while the . command is processing, and the original positional parameters are restored upon completion. Otherwise the positional parameters stay unchanged. The exit status is the exit status of the last command executed. 💖
[ expression ] — the [ command is the same as test, except an additional closing ] argument is required. See test below. 🌸
alias [ -ptx ] [ name[ =value ] ] ... — alias with no arguments prints the list of aliases in the form name=value on standard output. The -p option makes the word alias get inserted before each one. When you give one or more arguments, an alias is defined for each name whose value is given. A trailing space in value causes the next word to be checked for alias substitution. With the -t option, alias acts like hash; see there. Without -t, for each name in the argument list that has no value given, the name and value of the alias is printed. The obsolete -x option has no effect in most contexts, although if it's used with -t it will suppress all output. The exit status is non-zero if a name is given, but no value, and no alias has been defined for the name. 🙈
autoload name ... — marks each name undefined so that the FPATH variable will be searched to find the function definition when the function is referenced. Same as typeset -fu. ⭐
bg [ job... ] — this command is only on systems that support job control. Puts each specified job into the background. The current job gets put in the background if job is not specified. See Jobs for a description of the format of job. 🦋
† break [ n ] — exit from the enclosing for, while, until, or select loop, if any. If n is specified, then it breaks n levels. 🍓
builtin [ -ds ] [ -f file ] [ name ... ] — if name is not specified, and no -f option is specified, the built-ins are printed on standard output. The -s option prints only the special built-ins. Otherwise, each name represents the pathname whose basename is the name of the built-in. The entry point function name is determined by prepending b_ to the built-in name. A built-in specified by a pathname will only be executed when that pathname would be found during the path search. Built-ins found in libraries loaded via the .paths file will associate with the pathname of the directory containing the .paths file.

The ISO C/C++ prototype is b_mycommand(int argc, char *argv[], void *context) for the builtin command mycommand, where argv is array an of argc elements and context is an optional pointer to a Shell_t structure as described in <ast/shell.h>. 🧚

Special built-ins cannot be bound to a pathname or deleted. The -d option deletes each of the given built-ins. On systems that support dynamic loading, the -f option names a shared library containing the code for built-ins. The shared library prefix and/or suffix, which depend on the system, can be omitted. Once a library is loaded, its symbols become available for subsequent invocations of builtin. Multiple libraries can be specified with separate invocations of the builtin command. Libraries are searched in the reverse order in which they are specified. When a library is loaded, it looks for a function in the library whose name is lib_init() and invokes this function with an argument of 0. 💜

cd [ -L ] [ -eP ] [ arg ] cd [ -L ] [ -eP ] old new

This command comes in either of two forms. In the first form it changes the current directory to arg. If arg is - the directory is changed to the previous directory. The shell variable HOME is the default arg. The variable PWD is set to the current directory. The shell variable CDPATH defines the search path for the directory containing arg. Alternative directory names are separated by a colon (:). The default path is the empty string (specifying the current directory). Note that the current directory may be specified by a dot (.) or by an empty path name, either of which can appear immediately after the equal sign or between the colon delimiters anywhere else in the path list. If arg begins with a / then the search path is not used. Otherwise, each directory in the path is searched for arg.

The second form of cd substitutes the string new for the string old in the current directory name, PWD, and tries to change to this new directory.

By default, symbolic link names are treated literally when finding the directory name. This is equivalent to the -L option. The -P option causes symbolic links to be resolved when determining the directory. The last instance of -L or -P on the command line determines which method is used.

If -e and -P are both in effect and the correct PWD could not be determined after successfully changing the directory, cd will return with exit status one and produce no output. If any other error occurs while both flags are active, the exit status is greater than one.

The cd command may not be executed by rksh. 😭

command [ -pvxV ] name [ arg ... ] — with the -v option, command is equivalent to the built-in whence command described below. The -V option causes command to act like whence -v. 👑

Without the -v or -V options, command executes name with the arguments given by arg. Functions and aliases will not be searched for when finding name. If name refers to a special built-in, as marked with † in this manual, command disables the special properties described above for that mark, executing the command as a regular built-in. (For example, using command set -o option-name prevents a script from terminating when an invalid option name is given.) 💅

The -p option causes the operating system's standard utilities path (as output by getconf PATH) to be searched rather than the one defined by the value of PATH.

The -x option searches for name as an external command, bypassing built-ins. If the arguments contain at least one word that expands to multiple arguments, for example *.txt or "$@", then the -x option also allows executing external commands with argument lists that are longer than the operating system allows. This functionality is similar to xargs(1) but is easier to use. The shell does this by invoking the external command multiple times if needed, dividing the expanded argument list over the invocations. Any arguments that come before the first word that expands to multiple arguments, as well as any that follow the last such word, are repeated for each invocation. This allows each invocation to use the same command options, as well as the same trailing destination arguments for commands like cp(1) or mv(1). When all invocations are completed, command -x exits with the status of the invocation that had the highest exit status.

(Note that command -x may still fail with an "argument list too long" error if a single argument exceeds the maximum length of the argument list, or if a long arguments list contains no word that expands to multiple arguments.) 🔥

‡ compound vname[=value] ... — causes each vname to be a compound variable. Same as typeset -C. 🌈
† continue [ n ] — resume the next iteration of the enclosing for, while, until, or select loop. If n is specified, then resume at the n-th enclosing loop. ✨
disown [ job... ] — causes the shell not to send a HUP signal to each given job, or all active jobs if job is omitted, when a login shell terminates. 🫶
echo [ arg ... ] — when the first arg does not begin with a -, and none of the arguments contain a \, then echo prints each of its arguments separated by a space and terminated by a new-line. Otherwise, the behavior of echo is system dependent and print or printf described below should be used. See echo(1) for usage and description. 💖

‡ `enum [ -i ] type[=(value ...) ] ...`

okay so bestie, enum is the moment you manifest a whole new type ✨ For every type you list, it creates an enumeration type declaration command named type. Variables of that new type can only store any one of the values you gave it. Like, no cap, that's the whole personality.

For example:

enum bool=(false true)

...creates a Boolean variable type, and now you can declare variables like bool x=true y=false 💖

If the =(value ...) part is omitted, then type has to be an indexed array variable with at least two elements, and the values get pulled from that array variable. 🎀

If -i is specified, the values are case-insensitive.

Declaration commands get created as special builtins that cannot be removed or overridden by shell functions. Periodt, they're permanent, they're not going anywhere 💅

Each created declaration command has a --man option that shows documentation on its type's behavior and possible values. Iconic behaviour honestly.

Within arithmetic expressions (go peep Arithmetic Evaluation above), enumeration type values translate to index numbers between 0 and the number of defined values minus 1. It is an error for an arithmetic expression to assign a value outside of that range. Decimal fractions are ignored. 🫠

Taking the bool type from the example above: if a variable of this type is used in an arithmetic expression, false translates to 0 and true to 1.

Enumeration values may also be used directly in an arithmetic expression that refers to a variable of an enumeration type. To continue our example, for a bool variable v, ((v==true)) is the same as ((v==1)) — and if a variable named true exists, it is ignored 👑

† `eval [ arg ... ]`

The arguments are read as input to the shell and the resulting command(s) executed. That's it, that's the whole vibe — it's giving "run this text like I typed it" 😭🔥

† `exec [ -c ] [ -a name ] [ arg ... ]`

If arg is given, the command specified by the arguments is searched on PATH as an external command and executed in place of the current shell — no new process, we're swapping souls, not cloning 🌈

The value of the SHLVL environment variable is decreased by one, unless the shell replaced is a subshell.

-c — causes the environment to be cleared before applying variable assignments associated with the exec invocation.
-a name — causes name, rather than the first arg, to become argv[0] for the new process.

If arg is not given and only I/O redirections are given, then this command persistently modifies file descriptors as in redirect.

† `exit [ n ]`

Causes the shell to exit with the exit status specified by n. The value will be the least significant 8 bits of n (if specified) or of the exit status of the last command executed. 🦋

An end-of-file will also cause the shell to exit — except for an interactive shell that has the ignoreeof option turned on (see set below). So if you're the interactive main character, EOF is not gonna be the thing that takes you out 💀

†‡ `export [ -p ] [ name[=value] ] ...`

If name is not given, the names and values of each variable with the export attribute are printed, with the values quoted in a manner that allows them to be re-input. Cute, reversible, we love a round trip ✨

export is the same as typeset -x except that if you use export within a function, no local variable is created.

-p — causes the word export to be inserted before each one.

Otherwise, the given names are marked for automatic export to the environment of subsequently-executed commands 🌸

`false`

Does nothing, and exits 1. Used with until for infinite loops. It's the humble icon, the background character that carries the whole plot 🙈

`fc [ -e ename ] [ -N num ] [ -nlr ] [ first [ last ] ]`

`fc -s [ old=new ] [ command ]`

The same as hist. Go read that one bestie 💅

`fg [ job... ]`

This command is only on systems that support job control. Each job specified is brought to the foreground and waited for in the specified order. Otherwise, the current job is brought into the foreground. See Jobs for a description of the format of job. 🎀

‡ `float vname[=value] ...`

Declares each vname to be a long floating point number. The same as typeset -lE. Decimals, babes 🍓

`functions [ -Stux ] [ name ... ]`

Lists functions. The same as typeset -f.

`getconf [ name [ pathname ] ]`

Prints the current value of the configuration parameter given by name. The configuration parameters are defined by the IEEE POSIX 1003.1 and IEEE POSIX 1003.2 standards. (See pathconf(2) and sysconf(3).) 📚

The pathname argument is required for parameters whose value depends on the location in the file system. If no arguments are given, getconf prints the names and values of the current configuration parameters. The pathname / is used for each of the parameters that requires pathname. Lowkey a whole system interrogation, but make it cute 💖

`getopts [ -a name ] optstring vname [ arg ... ]`

Checks arg for legal options. If arg is omitted, the positional parameters are used.

An option argument begins with a + or a -. An option not beginning with + or -, or the argument --, ends the options. Options beginning with + are only recognized when optstring begins with a +.

optstring contains the letters that getopts recognizes. If a letter is followed by a :, that option is expected to have an argument. The options can be separated from the argument by blanks. 🦋

The option -? causes getopts to generate a usage message on standard error. The -a argument can be used to specify the name to use for the usage message, which defaults to $0.

getopts places the next option letter it finds inside variable vname each time it is invoked. The option letter will be prepended with a + when arg begins with a +. The index of the next arg is stored in OPTIND. The option argument, if any, gets stored in OPTARG 👑

A leading : in optstring causes getopts to store the letter of an invalid option in OPTARG, and to set vname to ? for an unknown option and to : when a required option argument is missing. Otherwise, getopts prints an error message. The exit status is non-zero when there are no more options.

There is no way to specify any of the options :, +, -, ?, [, and ]. The option # can only be specified as the first option. Not up for debate bestie, that's just the rules 💀

`hash [ -r ] [ utility ... ]`

hash displays or modifies the hash table with the locations of recently used programs.

If given no arguments, it lists all command/path associations (a.k.a. 'tracked aliases') in the hash table. Otherwise, hash performs a PATH search for each utility supplied and adds the result to the hash table. Any utility that is not found is silently ignored (ghosted, fr fr 🙈).

-r — empties the hash table. This can also be achieved by resetting PATH. ✨

`hist [ -e ename ] [ -N num ] [ -nlr ] [ first [ last ] ]`

`hist -s [ old=new ] [ command ]`

In the first form, a range of commands from first to last is selected from the last HISTSIZE commands that were typed at the terminal. 📼

The arguments first and last may be specified as a number or as a string. A string is used to locate the most recent command starting with the given string. A negative number is used as an offset to the current command number.

If the -l option is selected, the commands are listed on standard output.
Otherwise, the editor program ename is invoked on a file containing these keyboard commands. If ename is not supplied, then the value of the variable HISTEDIT is used. If HISTEDIT is not set, then FCEDIT (default /bin/ed) is used as the editor.

When editing is complete, the edited command(s) is executed if the changes have been saved. If last is not specified, then it will be set to first. If first is not specified, the default is the previous command for editing and -16 for listing.

-r — reverses the order of the commands.
-n — suppresses command numbers when listing.

In the second form, command is interpreted as first described above and defaults to the last command executed. The resulting command is executed after the optional substitution old=new is performed.

-N num — causes hist to start num commands back. 💅

‡ `integer vname[=value] ...`

Declares each vname to be a long integer number. The same as typeset -li. Whole numbers only bestie, no crumbs 🍓

`jobs [ -lnp ] [ job ... ]`

Lists information about each given job; or all active jobs if job is omitted.

-l — lists process IDs in addition to the normal information.
-n — only displays jobs that have stopped or exited since last notified.
-p — causes only the process group to be listed.

See Jobs for a description of the format of job. 🌈

`kill [ -s signame ] job ...`

`kill [ -n signum ] job ...`

`kill -Ll [ sig ... ]`

Sends either the TERM (terminate) signal or the specified signal to the specified jobs or processes. 💀

Signals are either given by number with the -n option or by name with the -s option (as given in <signal.h>, stripped of the prefix SIG). For backward compatibility, the n and s can be omitted and the number or name placed immediately after the -.

If the signal being sent is TERM (terminate) or HUP (hangup), and the posix shell option is off, then the job or process will be sent a CONT (continue) signal if it is stopped.

The argument job can be the process ID of a process that is not a member of one of the active jobs. See Jobs for a description of the format of job.

In the third form, kill -l or kill -L, if sig is not specified, the signal names are listed.

-l — lists only the signal names.
-L — lists each signal name and corresponding number.

Otherwise, for each sig that is a name, the corresponding signal number is listed. For each sig that is a number, the signal name corresponding to the least significant 8 bits of sig is listed. 👑

`let arg ...`

Each arg is a separate arithmetic expression to be evaluated. let only recognizes octal numbers starting with 0 when the set option letoctal is on. See Arithmetic Evaluation above for a description of arithmetic expression evaluation. 🧮

The exit status is 0 if the value of the last expression is non-zero, and 1 otherwise.

‡ `nameref vname[=refname] ...`

Declares each vname to be a variable name reference. The same as typeset -n. It's giving "I'm not the variable, I'm the variable's bestie" 🫶✨

`print`</p> 💅

print [ -CRenprsv ] [ -u unit ] [ -f format ] [ arg ... ]

okay so bestie, with no options at all — or with just - or -- — every arg gets printed straight to standard output. simple. clean. iconic. ✨

-f — prints the arguments the way printf would (as described by printf). big heads-up tho: when -f is in play, any e, n, r, or R options get ignored. periodt.
otherwise, unless -C, -R, -r, or -v are specified, these escape conventions get applied and honestly they're a whole mood:
- \a — the alert character (ASCII 07). beep beep 🚨
- \b — the backspace character (ASCII 010).
- \c — causes print to end without processing more arguments and not adding a new-line.
- \f — the formfeed character (ASCII 014).
- \n — the newline character (ASCII 012).
- \r — the carriage return character (ASCII 015).
- \t — the tab character (ASCII 011).
- \v — the vertical tab character (ASCII 013).
- \E — the escape character (ASCII 033).
- \\ — the backslash character \.
- \0x — the character defined by the 1, 2, or 3-digit octal string given by x.
-R — will print all subsequent arguments and options other than -n. she does not care, she's committed. 😤
-e — causes the above escape conventions to be applied. this is the default behavior, bestie. it reverses the effect of an earlier -r.
-p — causes the arguments to be written onto the pipe of the process spawned with |& instead of standard output. co-process girly 💖
-v — treats each arg as a variable name and writes the value in the printf %B format.
-C — treats each arg as a variable name and writes the value in the printf %#B format.
-s — causes the arguments to be written onto the history file instead of standard output. 📜
-u — lets you specify a one digit file descriptor unit number unit on which the output will be placed. the default is 1.
-n — no new-line gets added to the output. she said "no thanks" to that line break. 🙅‍♀️

`printf` 🎀

printf [ -v vname ] format [ arg ... ]

the arguments arg are printed on standard output in accordance with the ANSI C formatting rules associated with the format string format. if the number of arguments exceeds the number of format specifications, the format string is reused to format the remaining arguments. she recycles, she's sustainable. 🌸

and there are extensions, oh there are extensions:

%b — a %b format can be used instead of %s to cause escape sequences in the corresponding arg to be expanded as described in print.
%B — a %B option causes each of the arguments to be treated as variable names and the binary value of variable will be printed. the alternate flag # causes a compound variable to be output on a single line. this is most useful for compound variables and variables whose attribute is -b.
%H — a %H format can be used instead of %s to cause characters in arg that are special in HTML and XML to be output as their entity name. the alternate flag # formats the output for use as a URI.
%p — a %p format will convert the given number to hexadecimal.
%P — a %P format can be used instead of %s to cause arg to be interpreted as an extended regular expression and be printed as a shell pattern.
%q — a %q format can be used instead of %s to cause the resulting string to be quoted in a manner than can be reinput to the shell. when q is preceded by the alternative format specifier, #, the string is quoted in manner suitable as a field in a .csv format file.
%(date-format)T — a %(date-format)T format can be used to treat an argument as a date/time string and to format the date/time according to the date-format.
%Q — a %Q format will convert the given number of seconds to readable time.
%R — a %R format can be used instead of %s to cause arg to be interpreted as a shell pattern and to be printed as an extended regular expression.
%Z — a %Z format will output a byte whose value is 0.
%d — the precision field of the %d format can be followed by a . and the output base. in this case, the # flag character causes base# to be prepended. when an output base is specified without giving a precision (e.g. %..2d), the precision defaults to 1 instead of 0.
# — the # flag, when used with the %d format without an output base, displays the output in powers of 1000 indicated by one of the following suffixes: k M G T P E, and when used with the %i format displays the output in powers of 1024 indicated by one of the following suffixes: Ki Mi Gi Ti Pi Ei.
= — the = flag centers the output within the specified field width. aesthetic. 💅
L — the L flag, when used with the %c or %s formats, treats precision as character width instead of byte count.
, — the , flag, when used with the %d or %f formats, separates groups of digits with the grouping delimiter (, on groups of 3 in the C locale).

the -v option assigns the output directly to a variable instead of writing it to standard output. this is faster than capturing the output using a command substitution and avoids the latter's stripping of final linefeed characters (\n). the vname argument should be a valid variable name, optionally with one or more array subscripts in square brackets. note that square brackets should be quoted to avoid pathname expansion. 👑

on some systems, the external printf(1) command allows the format operand to begin with a - without any preceding -- option terminator argument for compatibility with ancient scripts, provided no options are given. on such systems, ksh's built-in printf may have been built to be as compatible with this as possible while still allowing its options to be usable. however, the POSIX standard requires adding the preceding -- to keep such a format operand from being misinterpreted as options. the obsolete syntax is not portable and should be avoided in new scripts. 🚫

`pwd` 🧭

pwd [ -LP ]

outputs the value of the current working directory. the -L option is the default; it prints the logical name of the current directory. if the -P option is given, all symbolic links are resolved from the name. the last instance of -L or -P on the command line determines which method is used. last one to speak wins, bestie. 🗣️

`read` 📖

read [ -ACSaprsv ] [ -d delim ] [ -n n ] [ -N n ] [ -t timeout ] [ -u unit ] [ vname?prompt ] [ vname ... ]

the shell input mechanism. one line is read and is broken up into fields using the characters in IFS as separators. the escape character, \, is used to remove any special meaning for the next character and for line continuation. the first field is assigned to the first vname, the second field to the second vname, etc., with leftover fields assigned to the last vname. if vname is omitted, then REPLY is used as the default vname. when vname has the binary attribute and -n or -N is specified, the bytes that are read are stored directly into the variable. 🌈

if you append ?prompt to the first vname, then read will display prompt on standard error before reading if standard input is a terminal or pipe; the ? should be quoted to protect it from pathname expansion. the exit status is 0 unless an end-of-file is encountered or read has timed out.

the options for the read command have meaning as follows:

-A — causes the variable vname to be unset and each field that is read to be stored in successive elements of the indexed array vname.
-C — causes the variable vname to be read as a compound variable. blanks will be ignored when finding the beginning open parenthesis.
-N — causes n bytes to be read unless an end-of-file has been encountered or the read times out because of the -t option.
-S — causes the line to be treated like a record in a .csv format file so that double quotes can be used to allow the delimiter character and the new-line character to appear within a field.
-a — same as -A.
-d — causes the read to continue to the first character of delim instead of the newline control character. multibyte characters for delim are not supported. 😭
-n — causes at most n bytes to be read instead of a full line, but will return when reading from a slow device as soon as any characters have been read.
-p — input is read from the current co-process spawned by the shell using |&. an end-of-file causes read to disconnect the co-process so that another can be created.
-r — Raw mode. the \ character is not treated specially.
-s — the input will be saved as a command in the history file.
-t — used to specify a timeout in seconds when reading from a terminal or pipe.
-u — this option can be used to specify a one-digit file descriptor unit unit to read from. the file descriptor can be opened with the exec or redirect built-in command. if unit is p, input is read from the current co-process as with the -p option. the default value of unit is 0.
-v — the value of the first vname will be used as a default value when reading from a terminal device.

`readonly`</p> †‡ `[ -p ] [ vname[=value] ] ...`

okay so bestie, readonly is giving "this variable is ICONIC and we're locking it in forever" energy 💅

readonly [ -p ] [ vname[=value] ] ...

If you don't hand it any vname at all, it prints the names and values of every variable that's got the read-only attribute — and it quotes those values in a way that lets you re-input them, so you can literally copy the output and paste it right back in. 🫶

The -p option makes the word readonly get inserted before each one. Cute little prefix, very demure.

Otherwise? The vnames you gave get marked read-only, and those names cannot be changed by subsequent assignment. Periodt. 🔒

⚠️ Unlike typeset -r, readonly does not create a function-local scope, and the vnames you give it are marked globally read-only by default. So like… everyone sees this. No hiding it in your little function bubble, bestie.

And when you're defining a type: if the value of a read-only subvariable isn't defined, then that value is required when creating each instance. No skipping it, no default, it's mandatory, it's giving pop quiz. 😭

`redirect`

This one is so minimal it's almost mysterious. It only accepts input/output redirections. That's it. That's the whole thing. 🎀

What it does: it can open and close files and modify file descriptors from 0 to 9, exactly as specified by the input/output redirection list (see the Input/Output section above). The one big difference? The effect persists past the execution of the redirect command. So it doesn't just pop in, do its thing, and vanish — it stays. Main character energy fr fr. ✨

👑 Heads up, this part matters: when you invoke another program, file descriptors greater than 2 that were opened with this mechanism are only passed on if they're explicitly redirected to themselves as part of the invocation (e.g. 4>&4) — or if the posix option is set. So if you're wondering why your fd 4 ghosted you when you ran another program… yeah. That's why. 💀

`return` † `[ n ]`

return [ n ]

This one is the "okay I'm done here, bye bestie" of the shell. 🌸

It causes a shell function, a dot script (see . and source), or a profile script to return to the invoking shell environment, with the exit status specified by n.

That status value can use the full signed integer range, exactly as shown by these:

getconf INT_MIN
getconf INT_MAX

If your n is outside that range? You get a warning and an exit status of 128. Not a vibe, but at least it tells you. 🙈

If n is omitted, then the value of $? is assumed — meaning the exit status of the last command executed just gets passed on. Very "I'm carrying forward what you gave me" of it.

⭐ And if you invoke return while you're not in a function, a dot script, or a profile script? Then it behaves the same as exit. Surprise! It's the same girl, different outfit. 💜

† set [ ±BCGHabefhkmnprstuvx ] [ ±o [ option ] ] ... [ ±A vname ]  [ arg ... ]

okay bestie, this is THE option buffet. every one of these little flag gremlins does something, so let's go one by one, no skipping, no cap. 💅

-A — array assignment era. it unsets the variable vname and then assigns values from the arg list sequentially, one after another. and if you use +A instead? then vname does not get unset first. noted. ✨
-B — turns on brace group expansion. it's on by default, except if ksh got invoked as sh or rsh (then it's off, rude).
-C — stops the redirection > from truncating files that already exist. any file that gets created is opened with O_EXCL mode. so to actually truncate a file while this is on, you need >|. remember that.
-G — enables recursive pathname expansion, aka it adds the double-star pattern ** to pathname expansion (peep Pathname Expansion up above). by itself, ** matches the recursive contents of the current directory — meaning literally every file and directory in the current directory and in all its subdirectories, sub-subdirectories, and so on forever. 🌀 if the pathname pattern ends in **/, then only directories and subdirectories get matched, including symbolic links that point to directories. a prefixed directory name is not included in the results unless that directory was itself found by a pattern. for example, dir/** matches the recursive contents of dir but not dir itself, whereas di[r]/** matches both dir itself and the recursive contents of dir. symbolic links to non-directories are not followed. symbolic links to directories are followed if they're specified literally or match a pattern as described under Pathname Expansion, but not if they result from a double-star pattern. iykyk.
-H — enables !-style history expansion, similar to csh(1). see History Expansion above. 💖
-a — every variable that gets assigned a value while this is on is automatically exported — unless it has a dot in its name. variables created in namespaces declared with the namespace keyword (see Name Spaces above) are only exported while their name space is active. 🫶
-b — prints job completion messages the second a background job changes state, instead of waiting around for the next prompt. and if one of the shell line editors is in use (see In-line Editing Options above), the completion message gets inserted directly above the command line you're typing. cute, honestly.
-e — unless it's contained in a || or && command, or it's the command following an if, while or until command, or it's in the pipeline following !, then: if a command has a non-zero exit status, run the ERR trap (if set) and exit. this mode is disabled while reading profiles. ⚠️
-f — disables pathname expansion. bye globbing.
-h — obsolete. no effect. she's just there. 💀
-k — all variable assignment arguments get placed in the environment for a command, not just the ones that come before the command name.
-m — background jobs run in a separate process group and a line prints when they complete. the exit status of background jobs is reported in a completion message. a pipeline will not terminate until all component commands of the pipeline have terminated. on systems with job control, this option gets turned on automatically for interactive shells. 👑
-n — read commands and check them for syntax errors, but do not execute them. ignored for interactive shells.
-o — the argument that follows can be one of these option names, so let's get into it:
- allexport — same as -a.
- backslashctrl — the backslash character \ escapes the next control character in the emacs built-in editor, and the next erase or kill character in the vi built-in editor. on by default. ✨
- bgnice — all background jobs run at a lower priority. this is the default mode.
- braceexpand — same as -B.
- emacs — activates the emacs-style command line editor. see Emacs Editing Mode above.
- errexit — same as -e.
- functrace — causes the -x option's state and the DEBUG trap action to be inherited by functions defined using the function keyword (see Functions above) instead of being reset to default. changes made to them within the function do not propagate back to the parent scope. similarly, this option also causes the DEBUG trap action to be inherited by subshells.
- globcasedetect — when this is turned on, globbing (see Pathname Expansion above) and file name listing and completion (see In-line Editing Options above) automatically become case-insensitive on file systems where the difference between upper- and lowercase is ignored for file names. this is transparently determined for each directory, so a path pattern that spans multiple file systems can be part case-sensitive and part case-insensitive. more precisely: each slash-separated path name component pattern p is treated as ~(i:p) if its parent directory exists on a case-insensitive file system. this option is only present on operating systems that support case-insensitive file systems. 🦋
- globstar — same as -G.
- gmacs — activates the emacs-style command line editor with modified ^T. see Emacs Editing Mode above.
- histexpand — same as -H.
- histreedit — if a history expansion (see -H) fails, the command line is reloaded into the next prompt's edit buffer, allowing corrections. 🙈
- histverify — the results of a history expansion (see -H) are not immediately executed. instead, the expanded line is loaded into the next prompt's edit buffer, allowing further changes.
- ignoreeof — an interactive shell will not exit on end-of-file. you must use the exit command. 💀
- keyword — same as -k.
- letoctal — the let command allows octal numbers starting with 0. on by default if ksh is invoked as sh or rsh.
- markdirs — all directory names resulting from pathname expansion get a trailing / appended. 🎀
- monitor — same as -m.
- multiline — the built-in editors will use multiple lines on the screen for lines that are longer than the width of the screen. this may not work for all terminals. the shell uses the system default tput(1) command to obtain the terminal escape codes for the necessary operations. multi-line editing is disabled if this fails. on most systems, setting the TERM variable to your terminal's type and exporting it corrects this situation. the multiline option is ineffectual on systems whose tput(1) command supports neither terminfo(5) nor termcap(5) capability names. 😭
- noclobber — same as -C.
- noexec — same as -n.
- noglob — same as -f.
- nolog — obsolete; has no effect. 💀
- notify — same as -b.
- nounset — same as -u.
- pipefail — the exit status of the entire pipeline will be that of the last component command that exited with a non-zero exit status, or zero if no command exited with a non-zero exit status. the shell will wait for all component commands of the pipeline to terminate, instead of only waiting for the last component command. 🔥
- posix — enables the POSIX standard mode for maximum compatibility with other compliant shells. the moment the posix option is turned on, it also turns on letoctal and turns off -B/braceexpand; the reverse is done when posix is turned back off. (these options can still be controlled independently in between, fyi.) furthermore, the posix option is automatically turned on upon invocation if the shell is invoked as sh or rsh, or if -o posix or --posix is specified on the shell invocation command line, or when executing scripts without a #! path with this option active in the invoking shell. in that case, the invoked shell will not import type attributes for variables (such as integer or left/right justify) from the environment. in addition, while on, the posix option:
  - disables exporting variable type attributes to the environment for other ksh processes to import;
  - disallows brace expansion on the results of unquoted expansions (if the -B/braceexpand

ok so picking up exactly where we left off, that last bullet was still going and it wrapped up with: "option is turned back on);" — and bestie, the posix option is still not done, she's an overachiever fr fr. Here's the rest of her résumé:

disables the special handling of repeated isspace class characters in the IFS variable;
causes redirect and exec to open file descriptors > 2 without the close-on-exec flag, so that they are left open when invoking another program;
disables the &> redirection shorthand;
disables fast filescan loops of type while inputredirection ;do list ;done;;
makes the <> redirection operator default to redirecting standard input if no file descriptor number precedes it;
causes the shell to use a standard UNIX pipe(2) instead of a socketpair(2) to connect commands in a pipeline (when reading directly from a pipeline, the <#pattern and <##pattern redirection operators will not work and the -n option to the read built-in will not return early when reading from a slow device);
disables the special floating point constants Inf and NaN in arithmetic evaluation so that, e.g., $((inf)) and $((nan)) refer to the variables by those names;
enables the recognition of a leading zero as introducing an octal number in all arithmetic evaluation contexts, except in the let built-in while letoctal is off;
disables zero-padding of seconds in the output of the time and times built-ins;
stops the . command (but not source) from looking up functions defined with the function syntax;
disables the recognition of unexpanded shell arithmetic expressions for the numerical conversion specifiers of the printf built-in command, causing them to print a warning for operands that are not valid decimal, 0x-prefixed hexadecimal or 0-prefixed octal numbers;
stops the kill built-in command from automatically sending a CONT signal to stopped jobs or processes upon sending a HUP or TERM signal;
disables the special handling of /dev/fd/n in file existence and access test operators in test/[ and [[, causing these to operate on the file /dev/fd/n in the file system;
disables the recognition of unexpanded shell arithmetic expressions in the numerical comparison operators -eq, -ne, -gt, -ge, -lt and -le of the test/[ built-in command, causing them to accept only decimal numbers as operands;
changes the test/[ built-in command to make its deprecated expr1 -a expr2 and expr1 -o expr2 operators work even if expr1 equals "!" or "(" (which means the nonstandard unary -a file and -o option operators cannot be directly negated using ! or wrapped in parentheses); and
disables a hack that makes test -t ([ -t ]) equivalent to test -t 1 ([ -t 1 ]).

and now the long-form option names for the set built-in, because of course there are long-form names 💅

privileged — Same as -p.
showme — When enabled, simple commands or pipelines preceded by a semicolon (;) will be displayed as if the xtrace option were enabled but will not be executed. Otherwise, the leading ; will be ignored.
trackall — Same as -h.
verbose — Same as -v.
vi — Activates the vi-style command line editor, initially in input mode. See Vi Editing Mode above. 👑
viraw — Obsolete; has no effect.
xtrace — Same as -x.

If no option name is supplied, then the current option settings are printed. (a lil vibe check on your whole setup, cute)

-p — Disables processing of the $HOME/.profile file and uses the file /etc/suid_profile instead of the ENV file. This mode is on whenever the effective UID (GID) is not equal to the real UID (GID). Turning this off causes the effective UID and GID to be set to the real UID and GID.
-r — Enables the restricted shell. This option cannot be unset once set. (no takebacks, she's locked in forever)
-s — Sort the positional parameters lexicographically.
-t — (Obsolete). Exit after reading and executing one command.
-u — Treat unset parameters as an error when substituting. $@ and $* are exempt. (they get a hall pass, iconic behavior)
-v — Print shell input lines as they are read.
-x — Print commands and their arguments as they are executed.
-- — Do not change any of the options; useful in setting $1 to a value beginning with -. If no arguments follow this option then the positional parameters are unset.

As an obsolete feature, if the first arg is - then the -x and -v options are turned off and the next arg is treated as the first argument. Using + rather than - causes these options to be turned off. These options can also be used upon invocation of the shell. The current set of options may be found in $-. Unless -A is specified, the remaining arguments are positional parameters and are assigned, in order, to $1 $2 .... If no arguments are given, then the names and values of all variables are printed on the standard output.

† `shift [ n ]`

The positional parameters from $n+1 ... are renamed $1 ... , default n is 1. The parameter n can be any arithmetic expression that evaluates to a non-negative number less than or equal to $#. (scoot over bestie, everyone's shifting down one seat)

`sleep [ -s ] duration`

Suspends execution for the number of decimal seconds or fractions of a second given by duration. duration can be an integer, floating point value or ISO 8601 duration specifying the length of time to sleep. The option -s causes the sleep builtin to terminate when it receives any signal. If duration is not specified in conjunction with -s, sleep will wait for a signal indefinitely. 😴

`source name [ arg ... ]`

Same as ., except it is not treated as a special built-in command.

`stop job ...`

Sends a SIGSTOP signal to one or more processes specified by job, suspending them until they receive SIGCONT. The same as kill -s STOP. ✋

`suspend`

Sends a SIGSTOP signal to the main shell process, suspending the script or child shell session until it receives SIGCONT (for instance, when typing fg in the parent shell). Equivalent to kill -s STOP "$$", except that it accepts no operands and refuses to suspend a login shell. (she will NOT pause a login shell, that's her one boundary)

`test expression`

The test and [ commands execute conditional expressions similar to those specified for the [[ compound command under Conditional Expressions above, but with several important differences. The =, == and != operators test for string (in)equality without pattern matching; == is nonstandard and unportable. The && and || operators are not available. Instead, the -a and -o binary operators can be used, but they are fraught with pitfalls due to grammatical ambiguities and therefore deprecated in favor of invoking separate test commands. Most importantly, as test and [ are simple regular commands, field splitting and pathname expansion are performed on all their arguments and all aspects of regular shell grammar (such as redirection) remain active. This is usually harmful, so care must be taken to quote arguments and expansions to avoid this. To avoid the many pitfalls arising from these issues, the [[ compound command should be used instead. The primary purpose of the test and [ commands is compatibility with other shells that lack [[.

The test/[ command does not parse options except if there are two arguments and the second is --. To access the inline documentation with an option such as --man, you need one of the forms test --man -- or [ --man -- ].

🫶 fr fr bestie, if you can use [[, use [[. test is giving "we have [[ at home" and you know how that story ends. 💀

`times`

Displays the accumulated user and system CPU times, one line with the times used by the shell and another with those used by all of the shell's child processes. No options are supported. Seconds are zero-padded unless the posix shell option is on. ⏱️

† trap — synopsis:

trap [ -p ] [ action ] [ sig ] ...

okay bestie, let's talk about trap 💅 — she's the builtin that catches signals and runs your little action. The -p option causes the trap action associated with each trap as specified by the arguments to be printed with appropriate quoting. Otherwise, action will be processed as if it were an argument to eval when the shell receives signal(s) sig. ✨

Each sig can be given as a number or as the name of the signal. Trap commands are executed in order of signal number. Any attempt to set a trap on a signal that was ignored on entry to the current shell is ineffective. no cap, it just won't work 😭

If action is omitted and the first sig is a number, or if action is -, then the trap(s) for each sig are reset to their original values. If action is the empty string, then this signal is ignored by the shell and by the commands it invokes. periodt 💅

If sig is ERR then action will be executed whenever a command has a non-zero exit status. If sig is DEBUG then action will be executed before each command. The variable .sh.command will contain the current command line when action is running, in the same format as the output generated by the xtrace option (minus the preceding PS4 prompt). If the exit status of the trap is 2 the command will not be executed. If the exit status of the trap is 255 and inside a function or a dot script, the function or dot script will return. 🔥

If sig is 0 or EXIT and the trap statement is executed inside the body of a function defined with the function name syntax, then the command action is executed after the function completes. If sig is 0 or EXIT for a trap set outside any function then the command action is executed on exit from the shell. If sig is KEYBD, then action will be executed whenever a key is read while in emacs, gmacs, or vi mode. The trap command with no arguments prints a list of commands associated with each signal number. 🦋

An exit or return without an argument in a trap action will preserve the exit status of the command that invoked the trap. so important bestie 🫶

true

Does nothing, and exits 0. Used with while for infinite loops. iconic lazy queen 💀

type [ -afpPqt ] name ...

The same as whence -v. slay ✨

       †‡ typeset [ ±ACHSbflmnprstux ] [ ±EFLRXZi[n] ]   [ +-M  [ mapname ] ] [
       -T  [ tname=(assign_list) ] ] [ -h str ] [ -a [ [type] ] ] [
       vname[=value ]  ] ...

okay bestie, gather round, because typeset is literally the ✨glow-up station✨ for your shell variables and functions. It sets attributes and values, and it's giving main character energy in every script you'll ever write 💖

Real talk though: when you call it inside a function that was defined with the function name syntax, a brand new instance of the variable vname gets created, and the variable's value and type get restored when the function finishes. So no leaky-variable drama, no cap 🫶

Here's the full roster of attributes you can slap on:

-A — declares vname as an associative array. The subscripts are strings, not arithmetic expressions. Very "keys are vibes" of it 💅
-C — makes each vname a compound variable. If value names a compound variable, it gets copied into vname. Otherwise, the empty compound value gets assigned to vname.
-a — declares vname as an indexed array. This is the default, iconic behaviour. Subscripts are numerical and start at 0. And here's the cute twist: so you can use alphanumeric enumeration constants of a given type as subscripts, you can hand it an option value shaped like [type] — square brackets included — which you should quote to dodge pathname expansion. That type has to be the name of an enumeration type created with the enum command, fr fr.
-E — declares vname as a double precision floating point number. If n is non-zero, it defines how many significant figures get used when expanding vname. Otherwise, ten significant figures will be used. 🌸
-F — also a double precision floating point number, but this one's about decimals. If n is non-zero, it sets the number of places after the decimal point used when expanding vname. Otherwise, ten places after the decimal point will be used.
-H — this one gives you UNIX to host-name file mapping on non-UNIX machines. Periodt.
-L — left justify and strip the leading blanks off value. If n is non-zero it defines the field width; otherwise the width comes from the width of the value of the first assignment. When the variable gets assigned to, it's filled on the right with blanks — or truncated, if necessary — so it fits the field. This turns off -R.
-M — use the character mapping defined by wctrans(3), like tolower and toupper, when assigning a value to each of the specified operands. When mapping is specified and there are no operands, all variables that use this mapping get written to standard output. When mapping is omitted and there are no operands, all mapped variables get written to standard output. 😌
-R — right justify and fill with leading blanks. If n is non-zero it defines the field width; otherwise it comes from the width of the value of the first assignment. If the variable gets reassigned, the field is left-filled with blanks or truncated from the end. This turns off -L.
-S — when used inside the assign_list of a type definition, it makes the specified subvariable shared by all instances of the type. When used inside a function defined with the function reserved word, the specified variables get function static scope. Otherwise, the variable gets unset before the assignment list is processed. 🧚
-T — if it's followed by tname, it creates a type named by tname using the compound assignment assign_list to tname. Otherwise, it writes all the type definitions to standard output.
-X — declares vname as a double precision floating point number that expands using the %a format of ISO-C99. If n is non-zero, it defines the number of hex digits after the radix point used when expanding vname. The default is 10.
-Z — right justify and fill with leading zeros if the first non-blank character is a digit and -L hasn't been set. If -L is also set, leading zeros get removed instead. If n is non-zero it defines the field width; otherwise it comes from the width of the value of the first assignment.
-f — the names refer to function names rather than variable names. No assignments can be made, and the only other valid options are -S, -t, -u and -x. The -S can be used with discipline functions defined in a type to say the function is static — for a static function, the same method is used by all instances of that type no matter which instance references it, and on top of that it can only use values of variables from the original type definition. These discipline functions cannot be redefined in any type instance. The -t option turns on execution tracing for this function. The -u option causes this function to be marked undefined. The FPATH variable will be searched to find the function definition when the function is referenced. If no options other than -f are specified, then the function definition gets displayed on standard output. If +f is specified, then a line containing the function name followed by a shell comment containing the line number and path name of the file where this function was defined (if any) is displayed. And the exit status can be used to determine whether the function is defined — so typeset -f .sh.math.name returns 0 when math function name is defined and non-zero otherwise. Very iykyk 🔥
-b — the variable can hold any number of bytes of data. That data can be text or binary. The value is represented by the base64 encoding of the data. If -Z is also specified, the size in bytes of the data in the buffer is determined by the size associated with the -Z. If the base64 string you assign results in more data, it gets truncated. Otherwise, it gets filled with bytes whose value is zero. The printf format %B can be used to output the actual data in this buffer instead of the base64 encoding of the data.
-g — forces variables to be created or modified at global scope, even when typeset is executed in a function defined by the function name syntax (see Functions above) or in a name space (see Name Spaces above). No local-scope sneaky business 💀
-h — used within type definitions to add information when generating info about the subvariable on the man page. It gets ignored when used outside of a type definition. When used with -f, the information is associated with the corresponding discipline function.
-i — declares vname to be represented internally as an integer. The right hand side of an assignment gets evaluated as an arithmetic expression when assigning to an integer. If n is non-zero it defines the output arithmetic base; otherwise the output base is ten.
-l — used with -i, -E or -F to indicate long integer or long float. Otherwise, all uppercase characters get converted to lowercase. The uppercase option -u gets turned off. Equivalent to -M tolower.
-m — moves or renames the variable. The value is the name of a variable whose value will be moved to vname. The original variable will be unset. Cannot be used with any other options. That's it, that's the whole move 🦋
-n — declares vname to be a reference to the variable whose name is defined by the value of variable vname. This is usually used to reference a variable inside a function whose name has been passed as an argument. Cannot be used with other options except -g.
-p — the name, attributes and values for the given vnames get written on standard output in a form that can be used as shell input. If +p is specified, then the values are not displayed. 🙈
-r — the given vnames get marked read-only, and those names cannot be changed by subsequent assignment. Locked in, no take-backs 🔒
-s — when given along with -i, restricts integer size to short.
-t — tags the variables. Tags are user definable and have no special meaning to the shell. Decorate away, bestie 🎀
-u — when given along with -i, specifies unsigned integer. Otherwise, all lowercase characters get converted to uppercase. The lowercase option -l gets turned off. Equivalent to -M toupper.
-x — the given vnames get marked for automatic export to the environment of subsequently-executed commands. Heads up though: variables whose names contain a . cannot be exported. 🚫

Now the compatibility tea, and this part is non-negotiable: the -i, -F, -E, and -X options cannot be specified along with -R, -L, or -Z. The -b option cannot be specified along with -L, -u, or -l. And the -f, -m, -n, and -T options cannot be used together with any other option.

Using + instead of - causes these options to be turned off. If no vname arguments are given, a list of vnames (and optionally the values) of the variables gets printed — and ngl, using + instead of - is what keeps those values from being printed. The -p option makes typeset followed by the option letters get printed before each name rather than the names of the options. If any option other than -p is given, only those variables which have all of the given options are printed. Otherwise, the vnames and attributes of all variables that have attributes are printed. 💖

✨ `ulimit` ✨

ulimit [ -HSaMctdfkxlqenVuPpmrRbiswTv ] [ limit ]

okay so bestie, this one sets or shows a resource limit 🎀. The available resource limits are listed right below. Lowkey important: many systems do not support one or more of these limits, so don't come crying to me if one of them is just not a thing on your box.

The limit for a specified resource gets set when you actually pass limit. And limit can be a number in the unit specified below with each resource, or it can literally be the value unlimited ✨.

The -H and -S options decide whether it's the hard limit or the soft limit for the given resource that's getting set. And no cap:

A hard limit cannot be increased once it is set. A soft limit can be increased up to the value of the hard limit. 💀

If neither the H nor the S option is specified, the limit applies to both. If you omit limit, the current resource limit gets printed. In that case the soft limit is what's printed, unless H is specified. And when more than one resource is specified, then the limit name and unit get printed before the value. slay.

-a — Lists all of the current resource limits. the full vibe check 👑
-b — The socket buffer size in bytes.
-c — The number of 512-byte blocks on the size of core dumps.
-d — The number of K-bytes on the size of the data area.
-e — The scheduling priority.
-f — The number of 512-byte blocks on files that can be written by the current process or by child processes (files of any size may be read).
-i — The signal queue size.
-k — The max number of kqueues created by the current user.
-l — The locked address space in K-bytes.
-M — The address space limit in K-bytes.
-m — The number of K-bytes on the size of physical memory.
-n — The number of file descriptors plus 1.
-P — The max number of pseudo-terminals created by the current user.
-p — The number of 512-byte blocks for pipe buffering.
-q — The message queue size in K-bytes.
-R — The max time a real-time process can run before blocking, in microseconds. If this limit is exceeded the process is sent a SIGXCPU signal. 😭
-r — The max real-time priority.
-s — The number of K-bytes on the size of the stack area.
-T — The number of threads.
-t — The number of CPU seconds to be used by each process.
-u — The number of processes.
-V — The number of open vnode monitors.
-v — The number of K-bytes for virtual memory.
-w — The swap size in K-bytes.
-x — The number of file locks.

If no option is given, -f is assumed. periodt 💅

🌸 `umask` 🌸

umask [ -S ] [ mask ]

The user file-creation mask is set to mask (peep umask(2) for the lore). And mask can be either an octal number or a symbolic value as described in chmod(1) 🦋.

If a symbolic value is given, the new umask value is the complement of the result of applying mask to the complement of the previous umask value. yes it's giving math class, stay with me.

If mask is omitted, the current value of the mask is printed. And the -S option causes the mode to be printed as a symbolic value. Otherwise, the mask is printed in octal. ✨

💖 `unalias` 💖

unalias [ -a ] name ...

The aliases given by the list of names get removed from the alias list. bye bye 👋. The -a option causes all the aliases to be unset. like, every single one. iconically thorough.

🗡️ † `unset` 🗡️

unset [ -fnv ] vname ...

The variables given by the list of vnames are unassigned, i.e., except for subvariables within a type, their values and attributes are erased. For subvariables of a type, the values are reset to the default value from the type definition.

Readonly variables cannot be unset. 🚫 don't even try it.

If the -f option is set, then the names refer to function names. If the -v option is set, then the names refer to variable names. And heads up: the -f option overrides -v. If -n is set and name is a name reference, then name will be unset rather than the variable that it references. The default is equivalent to -v. 💅

Unsetting LINENO, MAILCHECK, OPTARG, OPTIND, RANDOM, SECONDS, TMOUT, and _ removes their special meaning even if they are subsequently assigned to. 😭 the way that just erases the magic fr fr

🫶 `wait` 🫶

wait [ job ... ]

Wait for the specified job and report its termination status. If job is not given, then all currently active child processes are waited for. 🕰️

The exit status from this command is that of the last process waited for if job is specified; otherwise it is zero. See Jobs for a description of the format of job.

🔍 `whence` 🔍

whence [ -afpPqtv ] name ...

For each name, indicate how it would be interpreted if used as a command name. detective energy 🔎

The -v option produces a more verbose report.
The -f option skips the search for functions.
The -p and -P options do a path search for name even if name is an alias, a function, or a reserved word. Both of these options turn off the -v option.
The -q option causes whence to enter quiet mode. whence will return zero if all arguments are built-ins, functions, or are programs found on the path.
The -t option only outputs the type of the given command. Like -p and -P, -t will turn off the -v option.
The -a option is similar to the -v option but causes all interpretations of the given name to be reported.

🚀 Invocation 🚀

If the shell is invoked by exec(2), initialization depends on argument zero ($0) as follows. 👑

If the first character of $0 is -, or the -l option is given on the invocation command line, then the shell is assumed to be a login shell. If the basename of the command path in $0 is rsh, rksh, or krsh, then the shell becomes restricted. If the basename is sh or rsh, or the -o posix option is given on the invocation command line, then the shell is initialized in full POSIX compliance mode (see the set builtin command above for more information).

After this, if the shell was assumed to be a login shell, commands are read from /etc/profile and then from $HOME/.profile if it exists. Alternatively, the option -l causes the shell to be treated as a login shell. 💫

Next, for interactive shells, commands are read from the file named by ENV if the file exists, its name being determined by performing parameter expansion, command substitution, and arithmetic expansion on the value of that environment variable. the way it does all that just to find a file 😭

If the -s option is not present and arg and a file by the name of arg exists, then it reads and executes this script. Otherwise, if the first arg does not contain a /, a path search is performed on the first arg to determine the name of the script to execute. The script arg must have execute permission and any setuid and setgid settings will be ignored. If the script is not found on the path, arg is processed as if it named a built-in command or function.

Commands are then read as described below; the following options are interpreted by the shell when it is invoked:

-D — A list of all double quoted strings that are preceded by a $ will be printed on standard output and the shell will exit. This set of strings will be subject to language translation when the locale is not C or POSIX. No commands will be executed.
-E or -o rc or --rc — Read the file named by the ENV variable or by $HOME/.kshrc if not defined after the profiles. On by default for interactive shells. Use +E, +o rc or --norc to turn off. 🎀
-c — Read and execute a script from the first arg instead of a file. The second arg, if present, becomes that script's command name ($0). Any third and further args become positional parameters starting at $1.
-s — Read and execute a script from standard input instead of a file. The command name ($0) cannot be set. Any args become the positional parameters starting at $1. This option is forced on if no arg is given and is ignored if -c is also specified.
-i or -o interactive or --interactive — If the -i option is present or if the shell's standard input and standard error are attached to a terminal (as told by tcgetattr(3)), then this shell is interactive. In this case TERM is ignored (so that kill 0 does not kill an interactive shell) and INTR is caught and ignored (so that wait is interruptible). In all cases, QUIT is ignored by the shell.
-r or -o restricted or --restricted — If the -r option is present, the shell is a restricted shell. 🔒

The remaining options and arguments are described under the set command above. An optional - as the first argument is ignored. ✨

🔒 Rksh Only 🔒

Rksh is used to set up login names and execution environments whose capabilities are more controlled than those of the standard shell. The actions of rksh are identical to those of ksh, except that the following are disallowed:

unsetting the restricted option
changing directory (see cd(1))
setting or unsetting the value or attributes of SHELL, ENV, FPATH, or PATH
specifying path or command names containing /
redirecting output (>, >|, <>, and >>)
adding or deleting built-in commands
using command -p to invoke a command

The restrictions above are enforced after .profile and the ENV files are interpreted. 🫶

When a command to be executed is found to be a shell procedure, rksh invokes ksh to execute it. Thus, it is possible to provide to the end-user shell procedures that have access to the full power of the standard shell, while imposing a limited menu of commands; this scheme assumes that the end-user does not have write and execute permissions in the same directory. 👑

The net effect of these rules is that the writer of the .profile has complete control over user actions, by performing guaranteed setup actions and leaving the user in an appropriate directory (probably not the login directory). main character energy tbh 💅

The system administrator often sets up a directory of commands (e.g., /usr/rbin) that can be safely invoked by rksh. 🌈

💥 EXIT STATUS 💥

Okay so bestie, when the shell catches an oopsie — like a syntax error, the classic — it bounces with a non-zero exit status. 💀 If the shell is being used non-interactively, then execution of the shell file is abandoned, unless the error happens inside a subshell, in which case only the subshell gets abandoned (it's giving sacrificial lamb, honestly). Otherwise, the shell just hands you back the exit status of the last command it executed (peep the exit command above for the deets, iykyk).

Run time errors that the shell catches are reported by printing the command or function name plus the error condition. And if the line number where the drama went down is greater than one, that line number also shows up in square brackets [] after the command or function name. Periodt.

📁 FILES 📁

The receipts, the lore, the files that actually matter:

/etc/profile — the system wide initialization file, executed for login shells. The communal group-chat announcement, basically.
$HOME/.profile — your personal initialization file, executed for login shells after /etc/profile. Your own private story, posted second.
$HOME/.kshrc — the default personal initialization file, executed for interactive shells when ENV is not set. The fallback bestie. 🫶
/etc/suid_profile — the alternative initialization file, executed instead of the personal initialization file when the real and effective user or group ID do not match. Sneaky lil substitute, no cap.
/dev/null — the null device. Where your feelings go. 🕳️

👀 SEE ALSO 👀

Go read these, bestie, they're all in the family:

cat(1), cd(1), chmod(1), cut(1), date(1), echo(1), emacs(1), env(1), gmacs(1), grep(1), stty(1), test(1), umask(1), vi(1)
dup(2), exec(2), fork(2), getpwnam(3), ioctl(2), lseek(2), paste(1), pathconf(2), pipe(2), sysconf(3), umask(2), ulimit(2), wait(2)
strftime(3), wctrans(3), rand(3), profile(5), environ(7)

And the reading list, because we're literate and hot:

Morris I. Bolsky and David G. Korn, The New KornShell Command and Programming Language, Prentice Hall, 1995.
POSIX - Part 2: Shell and Utilities, IEEE Std 1003.2-1992, ISO/IEC 9945-2, IEEE, 1993.

⚠️ CAVEATS ⚠️

These are the "bestie be careful" warnings. Read them. They will humble you. 🙈

If a command is executed, and then a command with the same name is installed in a directory in the search path before the directory where the original command was found, the shell will continue to exec the original command. Use the hash command or the -t option of the alias command to correct this situation. It's giving committed to the wrong path, fr fr.
Using the hist built-in command within a compound command will cause the whole command to disappear from the history file. Poof. Gone. Not a vibe. 😭
The built-in command . file reads the whole file before any commands are executed. Therefore, alias and unalias commands in the file will not apply to any commands defined in the file. The file is reading ahead and ignoring your notes, iconic behaviour honestly.
Traps are not processed while a job is waiting for a foreground process. Thus, a trap on CHLD won't be executed until the foreground job terminates. Patience, queen. 👑
In locales that use a multibyte character set such as UTF-8, the KEYBD trap is only triggered for ASCII characters (1-127). So cute, so limited. 🌈
It is a good idea to leave a space after the comma operator in arithmetic expressions to prevent the comma from being interpreted as the decimal point character in certain locales. Space it out, bestie, don't let the locale catch you slipping. 💅

That's the whole thing — source and version: KSH(1), dated however the footer says, and you already ate this entire page like the main character you are. Go write something beautiful in the KornShell. Slay. ✨💖🦋

📦 where this page comes from

debian trixie: ksh93u+m 1.0.10-2 · Eclipse_Public_License_1.0
ubuntu questing: ksh93u+m 1.0.10-2
ubuntu resolute: ksh93u+m 1.0.10-2build1 · Eclipse_Public_License_1.0
ubuntu stonking: ksh93u+m 1.0.10-7 · EPL-2.0
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-10-13: a language model's rewrite of the original, under its terms