infocmp(1)

📦
version 6.6+20260608 · 8 versions
, in ubuntu stonking

💅 infocmp(1) 💅

okay bestie, welcome to infocmp — a User commands girly whose whole job is comparing and printing out terminfo descriptions. she's giving main character energy in the terminfo world, fr fr.

✨ NAME ✨

infocmp — compare or print out terminfo descriptions 🫶

🌸 SYNOPSIS 🌸

infocmp [-1cCdDeEFgGiIKlLnpqrtTuUVWx] [-A directory] [-B directory]
       [-Q encoding] [-R subset] [-s key] [-v level] [-w width]
       [terminal-type ... ]

🦋 DESCRIPTION 🦋

okay so infocmp does the most: it spits out a human-readable terminal type description from a compiled entry living in the terminfo database, and it can do that in a whole bunch of selectable formats. it also compares those entries against each other, and it can rewrite an entry so that any "use" expressions get swapped out with the actual content of other entries, by reference. iconic multitasking.

so a terminfo entry is basically three things holding hands: a list of one or more terminal type identifiers, a human-readable description of the terminal type, and a list of terminal capabilities that characterize its programming interface. periodt.

and in literally every case, the program reports Boolean-valued capabilities first, then the numeric ones, and then the string-valued capabilities. order is non-negotiable, she has a system. 💅

🎀 Default Options 🎀

if you specify no options and zero or one terminal-types, infocmp just assumes the -I option. no thoughts, head empty, terminfo codes.

if you specify more than one, the program assumes the -d option instead. she's flexible like that ✨

📚 Source Listing Options [-I] [-L] [-C] [-r] 📚

the -I, -L, and -C options will each produce a source listing for every terminal you name.

  • -I — use terminfo capability codes
  • -L — use "long" capability names
  • -C — use termcap capability codes
  • -r — with -C, include nonstandard capabilities
  • -K — with -C, improve BSD compatibility

if no terminal-types are given, the environment variable TERM gets used for the terminal name. she'll just guess, it's fine 💖

the source produced by the -C option can be used directly as a termcap entry, but bestie, not all parameterized strings can be changed to termcap format. infocmp will attempt to convert most of the parameterized information, and anything it doesn't convert will be plainly marked in the output and commented out. those ones? you gotta edit by hand. sorry queen 😭

for best results when converting to termcap format, use both -C and -r. normally a termcap description is limited to 1023 bytes, and infocmp trims away less essential parts to make it fit. if you're converting to one of the (rare) termcap implementations that accept an unlimited size of termcap, you might wanna add the -T option. more often though, you have to help the termcap implementation out and trim excess whitespace — use the -0 option for that ✨

all padding information for strings gets collected together and placed at the beginning of the string, which is where termcap expects it. and mandatory padding (that's padding information with a trailing /) will become optional. casual.

all termcap variables that terminfo no longer supports — but which are derivable from other terminfo variables — will be output. not all terminfo capabilities get translated; only those variables that were part of termcap will normally be output. specifying the -r option takes off that restriction, allowing all capabilities to be output in termcap form. normally you'd use both -C and -r together, like a matching set 🫶

the actual format used incorporates some improvements for escaped characters from terminfo format. for a stricter BSD-compatible translation, use the -K option rather than -C.

note that because padding is collected to the beginning of the capability, not all capabilities are output. and mandatory padding is not supported. because termcap strings are not as flexible, it is not always possible to convert a terminfo string capability into an equivalent termcap format. a subsequent conversion of the termcap file back into terminfo format will not necessarily reproduce the original terminfo source. 💀

some common terminfo parameter sequences, their termcap equivalents, and some terminal types which commonly have such sequences, are:

terminfo                   termcap   Terminal Types
────────────────────────────────────────────────────
%p1%c                      %.        ansi-m
%p1%d                      %d        ansi, vt100
%p1%' '%+%c                %+x       vt52
%i                         %iq       ansi, vt100
%p1%?%'x'%>%t%p1%'y'%+%;   %>xy      annarbor4080
%p2...%p1                  %r        hpgeneric

🔍 Entry Comparison Options [-d] [-c] [-n] 🔍

given -c, -d, or -n, infocmp compares the terminfo description of the first specified terminal-type with those of each of the subsequent operands. if fewer terminal-types than required are specified, infocmp uses the environment variable TERM in their place. she makes it work 💅

if a capability is defined for only one terminal type, the value reported depends on the capability's type:

  • F for missing Boolean variables
  • NULL for missing integer or string variables

the -c and -d options report string capability values between ' characters. use the -q option to distinguish absent and canceled capabilities; see terminfo(5).

the comparison option selects the form of report:

  • -d — lists each capability that differs between two entries. each capability name is followed by : and comma-separated capability values, then a period.
  • -c — lists each capability that two entries have in common. infocmp ignores capabilities missing from either entry. each capability name is followed by =, a space, and the capability value, then a period.
    • if the -u option is further specified, infocmp rewrites the description of the first type employing use= syntax to use the second as a building block.
  • -n — lists capabilities that are in none of the given entries. each capability name is preceded by ! and followed by a period.
    • normally only conventional capabilities are shown. use the -x option to add BSD-compatibility capabilities (names prefixed with OT).

💖 Use= Option [-u] 💖

the -u option produces a terminfo source description of the first terminal terminal-type which is relative to the sum of the descriptions given by the entries for the other terminal-types. she does this by analyzing the differences between the first terminal-types and the other terminal-types and producing a description with use= fields for the other terminals. in this manner, it's possible to retrofit generic terminfo entries into a terminal's description. or, if two similar terminals exist but were coded at different times or by different people so that each description is a full description, using infocmp will show you what can be done to change one description to be relative to the other. glow-up for your config 🦋

a capability will be printed with an at-sign (@) if it no longer exists in the first terminal-type, but one of the other terminal-type entries contains a value for it. a capability's value will be printed if the value in the first terminal-type is not found in any of the other terminal-type entries, or if the first of the other terminal-type entries that has this capability gives a different value for the capability than that in the first terminal-type.

and the order of the other terminal-type entries is significant, no cap. since the terminfo compiler tic does a left-to-right scan of the capabilities, specifying two use= entries that contain differing entries for the same capabilities will produce different results depending on the order that the entries are given in. infocmp will flag any such inconsistencies between the other terminal-type entries as they are found 👑

alternatively, specifying a capability after a use= entry that contains that capability will cause the second specification to be ignored. using infocmp to recreate a description can be a useful check to make sure everything was specified correctly in the original source description. self-care tbh.

another error that does not cause incorrect compiled files, but will slow down the compilation time, is specifying extra use= fields that are superfluous. infocmp will flag any other terminal-type use= fields that were not needed. she notices everything ✨

🗂️ Changing Databases [-A directory] [-B directory] 🗂️

like other ncurses utilities, infocmp looks for the terminal descriptions in several places. you can use the TERMINFO and TERMINFO_DIRS environment variables to override the compiled-in default list of places to search. see ncurses(3NCURSES), as well as the Fetching Compiled Descriptions section in terminfo(5).

you can also use the options -A and -B to override the list of places to search when comparing terminal descriptions:

  • the -A option sets the location for the first terminal-type
  • the -B option sets the location for the other terminal-types

using these options, it's possible to compare descriptions for a terminal with the same name located in two different databases. for instance, you can use this feature for comparing descriptions for the same terminal created by different people. detective work but make it cute 🔍💅

🍓 Other Options 🍓

  • -0 — causes the fields to be printed on one line, without wrapping.
  • -1 — causes the fields to be printed out one to a line. otherwise, the fields will be printed several to a line to a maximum width of 60 characters.
  • -a — tells infocmp to retain commented-out capabilities rather than discarding them. capabilities are commented by prefixing them with a period.
  • -D — tells infocmp to print the database locations that it knows about, and exit.
  • -E — dump the capabilities of the given terminal as tables, needed in the C initializer for a TERMTYPE structure (the terminal capability structure in the <term.h>). this option is useful for preparing versions of the curses library hardwired for a given terminal type. the tables are all declared static, and are named according to the type and the name of the corresponding terminal entry.
    • before ncurses 5.0, the split between the -e and -E options was not needed; but support for extended names required making the arrays of terminal capabilities separate from the TERMTYPE structure. lore 🫶
  • -e — dump the capabilities of the given terminal as a C initializer for a TERMTYPE structure (the terminal capability structure in the <term.h>). this option is useful for preparing versions of the curses library hardwired for a given terminal type.

more coming bestie, don't go anywhere 💖

  • -F — bestie, this one is a comparison mode. It assumes the next two arguments are filenames. The files get searched for pairwise matches between entries, and two entries count as a match if any of their names match. The report goes to standard output and lists entries with no matches in the other file, plus entries with more than one match. For entries with exactly one match, it also includes a difference report. Normally, to keep the report from getting huge, use references are not resolved before looking for differences — but you can force that resolution by also specifying -r. 💅
  • -f — display complex terminfo strings that contain if/then/else/endif expressions, indented so it's actually readable. 🦋
  • -G — display constant literals in decimal form instead of their character equivalents.
  • -g — display constant character literals in quoted form instead of their decimal equivalents. (Yes, it's the opposite of -G, they're a matched set, iconic.)
  • -i — analyze the initialization strings (is1, is2, is3) and the reset strings (rs1, rs2, rs3) in the entry, plus the ones used for starting/stopping cursor-positioning mode (smcup, rmcup) and for starting/stopping keymap mode (smkx, rmkx). 🧚

For each string, the code tries to break it down into actions in terms of

  • the other capabilities in the entry,
  • certain X3.64/ISO 6429/ECMA-48 capabilities,
  • certain DEC VT-series private modes, and
  • certain XTerm control sequences.

The set of recognized special sequences was picked for completeness over the existing terminfo database. Each report line is the capability name, then a colon and a space, then a printable expansion of the capability string where the sections matching recognized actions get translated into {}-bracketed descriptions.

Here's the whole list of special sequences it recognizes, bestie:

  • ED — clear screen
  • ED2 — clear page
  • HOME — move cursor to home
  • HTS — horizontal tag set
  • LL — home-down
  • NEL — next line
  • RC — restore cursor
  • RI — reverse index
  • RIS — full reset
  • RSR — reset scroll region
  • SC — save cursor
  • SGR0 — reset video attributes
  • SI — shift-in
  • SO — shift-out
  • DECSCL — set conformance level (VT220)
  • DECSTR — soft reset (VT320)
  • S7C1T — 7-bit controls (VT220)
  • S8C1T — 8-bit controls (VT220)
  • ISO DEC G0 — enable DEC graphics for G0
  • ISO UK G0 — enable UK chars for G0
  • ISO US G0 — enable US chars for G0
  • ISO DEC G1 — enable DEC graphics for G1
  • ISO UK G1 — enable UK chars for G1
  • ISO US G1 — enable US chars for G1
  • DECPAM — application keypad mode
  • DECPNM — normal keypad mode
  • DECANSI — enter ANSI mode
  • ECMA[+-]GATM — guarded area transfer mode
  • ECMA[+-]AM — keyboard action mode
  • ECMA[+-]CRM — control representation mode
  • ECMA[+-]IRM — insert replace mode
  • ECMA[+-]STRM — status report transfer mode
  • ECMA[+-]ERM — erasure mode
  • ECMA[+-]SRM — send receive mode
  • ECMA[+-]FEAM — format effector action mode
  • ECMA[+-]LNM — linefeed mode
  • DEC[+-]CKM — application cursor keys
  • DEC[+-]ANM — set VT52 mode
  • DEC[+-]COLM — 132-column mode
  • DEC[+-]SCLM — smooth scroll
  • DEC[+-]SCNM — reverse video mode
  • DEC[+-]OM — origin mode
  • DEC[+-]AWM — wraparound mode
  • DEC[+-]ARM — auto-repeat mode
  • DEC[+-]TCEM — text cursor enable
  • DEC[+-]NRCM — national replacement character mode
  • DEC[+-]BKM — backarrow key mode
  • DEC[+-]X11MSE — X11 mouse (XTerm)
  • DEC[+-]META — meta/eightBitInput (XTerm)
  • DEC[+-]ALTSCRN — alternate screen (XTerm)
  • push title — save icon and window title (XTerm)
  • pop title — restore icon and window title (XTerm)

It also recognizes an SGR action corresponding to ANSI/ISO 6429/ECMA Set Graphics Rendition, with the values NORMAL, BOLD, UNDERLINE, BLINK, and REVERSE. All of them except NORMAL can be prefixed with

  • “+” (turn on) or
  • “-” (turn off).

An SGR0 designates an empty highlight sequence (equivalent to {SGR:NORMAL}). No notes, that's just how it is. ✅

  • -l — set output format to terminfo.
  • -p — ignore padding specifications when comparing strings.
  • -Q n — instead of showing source in terminfo (text) format, print the compiled (binary) format in hexadecimal or base64 form, depending on the option's value:
    • 1 — hexadecimal
    • 2 — base64
    • 3 — hexadecimal and base64

For example, this prints the compiled terminfo value as a string that could be assigned to the TERMINFO environment variable:

infocmp -0 -q -Q2
  • -q — this makes the output a little shorter:
    • Make the comparison listing shorter by omitting subheadings, and using “-” for absent capabilities and “@” for canceled rather than “NULL”.
    • However, show differences between absent and canceled capabilities.
    • Omit the “Reconstructed from” comment for source listings.
  • -Rsubset — restrict output to a given subset. This option is for use with archaic versions of terminfo like those on SVr1, Ultrix, or HP-UX that don't support the full set of SVR4/XSI Curses terminfo, and variants such as AIX that have their own extensions incompatible with SVr4/XSI.
    • Available terminfo subsets are “SVr1”, “Ultrix”, “HP”, and “AIX”; see terminfo(5) for details.
    • You can also choose the subset “BSD” which selects only capabilities with termcap equivalents recognized by 4.4BSD.
    • If you select any other value for -R, it's the same as no subset, i.e., all capabilities are used.

A few options override the subset selected with -R, if they get processed later in the command parameters:

  • -C sets the “BSD” subset as a side effect.
  • -I sets the subset to all capabilities.
  • -r sets the subset to all capabilities.
  • -s [d|i|l|c] — the -s option sorts the fields within each type according to the argument below:
    • d — leave fields in the order that they are stored in the terminfo database.
    • i — sort by terminfo name.
    • l — sort by the long C variable name.
    • c — sort by the termcap name.

If the -s option isn't given, the fields printed out will be sorted alphabetically by the terminfo name within each type — except in the case of the -C or the -L options, which cause the sorting to be done by the termcap name or the long C variable name, respectively.

  • -T — eliminates size-restrictions on the generated text. This is mainly useful for testing and analysis, since the compiled descriptions are limited (e.g., 1023 for termcap, 4096 for terminfo).
  • -t — tells tic to discard commented-out capabilities. Normally when translating from terminfo to termcap, untranslatable capabilities are commented-out.
  • -U — tells infocmp to not post-process the data after parsing the source file. This feature helps when comparing the actual contents of two source files, since it excludes the inferences that infocmp makes to fill in missing data. 🙈
  • -V — reports the version of ncurses which was used in this program, and exits.
  • -v n — prints out tracing information on standard error as the program runs.

The optional parameter n is a number from 1 to 10, inclusive, indicating the desired level of detail of information. If ncurses is built without tracing support, the optional parameter is ignored.

  • -W — By itself, the -w option will not force long strings to be wrapped. Use the -W option to do this. 🌸
  • -w width — changes the output to width characters.
  • -x — print information for user-defined capabilities (see user_caps(5).) These are extensions to the terminfo repertoire which can be loaded using the -x option of tic.

📂 FILES 📂

  • /etc/terminfo — compiled terminal description database

🌟 EXTENSIONS 🌟

The -0, -1, -a, -e, -E, -f, -F, -g, -G, -i, -l, -p, -q, -Q, -R, -t, -T, and -V options are ncurses extensions. ✨ (yes that's a lot of extensions, we contain multitudes)

🛂 PORTABILITY 🛂

X/Open Curses Issue 7 (2009) specifies infocmp. It does not mention options for producing descriptions in termcap format.

SVr4 infocmp does not distinguish between absent and canceled capabilities. It furthermore reports missing integer capabilities as -1 (its internal representation). ncurses shows these as “NULL” for consistency with missing string capabilities.

The -r option of ncurses infocmp uses SVr4's notion of “termcap” capabilities. BSD curses had a more restricted set. To see only those present in 4.4BSD, use “-r -RBSD”. 💜

📜 HISTORY (storytime, besties) 📜

So back in the day, System V Release 2 rolled up with a terminfo library and was like "I'm here 💅" — except it had no documented tool for decompiling the terminal descriptions. Like, the vibes were immaculate but the receipts? Nowhere. 😭

Then Tony Hansen (from AT&T, king behavior 👑) wrote the first infocmp in early 1984, for System V Release 3. Iconic. Periodt. ✨

Fast forward to 1995 and Eric Raymond took the AT&T documentation and used it to make an equivalent infocmp for ncurses. Not only that, he added a few new features, because why not slay:

  • the -e option, to support fallback (compiled-in) terminal descriptions 🫶
  • the -i option, to help with analysis 🔍

Later on, Thomas Dickey pulled up and added the -x (user-defined capabilities) option, plus the -E option to support fallback entries with user-defined capabilities. Two for two, no notes. 🔥

For a complete list, go look at the EXTENSIONS section, bestie — it's all in there. 👀

And in 2010, Roy Marples provided an infocmp program for NetBSD. Ngl it is less capable than the SVr4 or ncurses versions (e.g., it lacks the sorting options documented in X/Open 😔), but it does include the -x option adapted from ncurses. So like... it showed up, it did its thing, we respect it. 🎀

💀 BUGS 💀

The -F option of infocmp(1) should be a toe(1) mode.

That's it. That's the bug. Not me exposing the whole thing. 🙈

👑 AUTHORS 👑

The legends themselves:

  • Eric S. Raymond <esr@snark.thyrsus.com>
  • Thomas E. Dickey <dickey@invisible-island.net>

Give them their flowers. 💐

🌸 SEE ALSO 🌸

Go read these next, iykyk:

And this lil link if you wanna go deeper:

https://invisible-island.net/ncurses/tctest.html

okay that's the whole page, we ate and left no crumbs 💅✨

This little gem comes from ncurses 6.6, dated 2026-02-07, and it's the infocmp(1) girlies' own manual. Now go decompile something cute, bestie. 🦋💖🌸

📦 where this page comes from