Algori...nential(3pm)

📦
version 0.010 · 5 versions
, in debian trixie, ubuntu resolute, ubuntu stonking

💖 NAME 💖

Algorithm::Backoff::Exponential — bestie, it backs off exponentially. That's the whole vibe. It's a Perl module (you'll find it filed under 3pm, the User Contributed Perl Documentation manual, which is why the header is all squished like Algori...nential(3pm) — the terminal ate the middle of her name 😭).

🎀 VERSION 🎀

This doc is describing version 0.010 of Algorithm::Backoff::Exponential, which ships in the Perl distribution Algorithm-Backoff, released on 2024-02-24. Yes she's got a birthday, and yes she's thriving. 🎂

✨ SYNOPSIS ✨

Okay so first you instantiate her, then you feed her successes and failures and she hands you back a number of seconds to wait. Like a little delay oracle. 🔮

use Algorithm::Backoff::Exponential;

# 1. instantiate

my $ab = Algorithm::Backoff::Exponential->new(
    #consider_actual_delay => 1, # optional, default 0
    #max_actual_duration   => 0, # optional, default 0 (retry endlessly)
    #max_attempts          => 0, # optional, default 0 (retry endlessly)
    #jitter_factor         => 0.25, # optional, default 0
    initial_delay          => 5, # required
    #max_delay             => 100, # optional
    #exponent_base         => 2, # optional, default 2 (binary exponentiation)
    #delay_on_success      => 0, # optional, default 0
);

# 2. log success/failure and get a new number of seconds to delay, timestamp is
# optional but must be monotonically increasing.

# for example, using the parameters initial_delay=5, max_delay=100:

my $secs;
$secs = $ab->failure();   # =>  5 (= initial_delay)
$secs = $ab->failure();   # => 10 (5 * 2^1)
$secs = $ab->failure();   # => 20 (5 * 2^2)
$secs = $ab->failure();   # => 33 (5 * 2^3 - 7)
$secs = $ab->failure();   # => 80 (5 * 2^4)
$secs = $ab->failure();   # => 100 ( min(5 * 2^5, 100) )
$secs = $ab->success();   # => 0 (= delay_on_success)

And here's her serving looks from the command line via show-backoff-delays — that's 10 failures followed by 3 successes:

% show-backoff-delays -a Exponential --initial-delay 1 --max-delay 200 \
    0 0 0 0 0   0 0 0 0 0   1 1 1
1
2
4
8
16
32
64
128
200
200
0
0
0

Look at her go. 1, 2, 4, 8, 16, 32, 64, 128, then she hits the 200 ceiling twice and just chills there, and the second a success rolls in she's back at 0. Character development. 💅

🌸 DESCRIPTION 🌸

So the way this backoff algorithm does math is:

initial_delay * exponent_base ** (attempts-1)

Only initial_delay is required — everything else is optional, she's low maintenance like that. exponent_base defaults to 2 (that's the binary exponential situation). On the very first failure attempt (when attempts = 1) the delay is just the initial delay, and from there she doubles, quadruples, and keeps on climbing (with that default exponent base of 2, obviously).

There are ceilings, bestie — limits on how many attempts you get (max_attempts) and how long the whole thing can drag on (max_actual_duration).

Pro tip from the docs themselves: add a jitter factor, like 0.25, to sprinkle in some randomness and dodge the "thundering herd problem". 🐘🐘🐘

🔧 METHODS 🔧

new

Usage, exactly as she comes:

new(%args) -> obj

This function is not exported. So you're calling it as a method, not importing it. Noted. 📝

The arguments, where * means required (no skipsies):

  • consider_actual_delay => bool (default: 0) — whether to consider actual delay.

If you set this one to true, she takes the actual delay into account (that's the timestamp difference). Example time: with the Constant strategy of delay=2, if you call failure() again right after the previous failure() (same timestamp, no shame), failure() returns roughly 2+2 = 4 seconds. But if you actually waited 2 seconds before calling failure() again (timestamp 2 seconds bigger than the last one), failure() returns 2 seconds. And if you waited 4 seconds or more? failure() hands you 0. Iconic behavior honestly. ✨

  • delay_on_success => ufloat (default: 0) — number of seconds to wait after a success.
  • exponent_base => ufloat (default: 2) — (No description). She's mysterious, she's just vibing, the docs say nothing and that's fine. 🙈
  • initial_delay* => ufloat — the initial delay for the first attempt after a failure, in seconds. This one's required, bestie, don't forget her.
  • jitter_factor => float — how much randomness to add.

If you set this to anything larger than 0, the actual delay lands somewhere in a random number between original_delay * (1-jitter_factor) and original_delay * (1+jitter_factor). Jitters get added to avoid the so-called "thundering herd" problem. And this jitter applies to the delay on failure as well as on success. Equal opportunity jitter. 🎲

  • max_actual_duration => ufloat (default: 0) — maximum number of seconds for all of the attempts (0 means unlimited).

If you set it to a positive number, she caps how many seconds the whole set of attempts can take. It's how you say "I'm only willing to spend this much time on this task." Example: with the Exponential strategy of initial_delay=3 and max_attempts=10, the delays go 3, 6, 12, 24, .... If failures get logged along those suggested delays and max_actual_duration is set to 21 seconds, then the third failure() returns -1 instead of 24, because 3+6+12 >= 21 — even though max_attempts was never exceeded. She said "we're done here." 💀

  • max_attempts => uint (default: 0) — maximum number of consecutive failures before giving up.

0 means retry endlessly, forever, never giving up, the grind never stops. 1 means give up after a single failure (so, zero retry attempts). 2 means retry once after a failure. And heads up: after a success the attempt count resets (as you'd expect). So if max_attempts is 3 and you fail twice and then succeed, then on the next failure she starts the whole retry thing over again for up to 3 times. Fresh start, new era. 🌈

  • max_delay => ufloat — maximum delay time, in seconds.
  • min_delay => ufloat (default: 0) — maximum delay time, in seconds. Yep, that's literally what the docs say, word for word — "Maximum delay time, in seconds." on the min_delay entry too. I'm not touching it, I'm just reporting. 🫶

Return value: (obj)

🦋 HOMEPAGE 🦋

Please go visit the project's homepage, she'd love that:

<https://metacpan.org/release/Algorithm-Backoff>

⭐ SOURCE ⭐

The source repository lives at:

<https://github.com/perlancar/perl-Algorithm-Backoff>

👑 SEE ALSO 👑

💜 AUTHOR 💜

perlancar <perlancar@cpan.org>

🧚 CONTRIBUTING 🧚

Wanna contribute? You can send patches by email / via RT, or send pull requests on GitHub. She welcomes you with open arms. 🫶

Most of the time you do not need to build the distribution yourself. You can just modify the code and then test it like:

% prove -l

If you do want to build the distribution (say, to try installing it locally on your system), you can install Dist::Zilla, Dist::Zilla::PluginBundle::Author::PERLANCAR, Pod::Weaver::PluginBundle::Author::PERLANCAR, and sometimes one or two other Dist::Zilla- and/or Pod::Weaver plugins. Any extra steps required beyond that are considered a bug and you can report them. No cap.

This software is copyright (c) 2024, 2019 by perlancar <perlancar@cpan.org>.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself. Slay, legally. ⚖️✨

💀 BUGS 💀

Please report any bugs or feature requests on the bugtracker website:

<https://rt.cpan.org/Public/Dist/Display.html?Name=Algorithm-Backoff>

And when you submit a bug or a request, please include a test-file or a patch to an existing test-file that shows off the bug or the desired feature. Help her help you, bestie. 🐛

And that's the whole page, mwah 💅 — this was Algorithm::Backoff::Exponential, served to you from perl v5.40.1 on 2025-05-04, and that header up top that looked like Algori...nential(3pm) was just her name with the middle cut out, but we all know who she is. 🎀 Go forth and back off exponentially. 💖

📦 where this page comes from