[prev in list] [next in list] [prev in thread] [next in thread] 

List:       busybox
Subject:    Re: ash ${str/find/repl} performance
From:       Alin Mr <almr.oss () outlook ! com>
Date:       2021-07-21 21:14:38
Message-ID: AM8PR08MB5795A28D22DA78B371EFEFAEF7E39 () AM8PR08MB5795 ! eurprd08 ! prod ! outlook ! com
[Download RAW message or body]

Bernd,

> busybox is about size, not speed.

Yes, I know busybox from embedded work. Surprisingly, switching to busybox in a \
"normal" Linux app (Kakoune -- a modal editor that uses POSIX shell as an extension \
language) made the program a lot faster. Like, twice faster for startup:

Summary
  'KKPSH=bboxsh/builtin+nofork/sh kakrun' ran
    1.07 ± 0.02 times faster than 'KKPSH=bboxsh/builtin/sh kakrun'
    1.21 ± 0.01 times faster than 'KKPSH=bboxsh/links/sh PATH="bboxsh/links:$PATH" \
kakrun'  1.40 ± 0.01 times faster than 'KKPSH=bboxsh/builtin+dynld/sh kakrun'
    2.08 ± 0.03 times faster than 'KKPSH=bboxsh/links/sh kakrun'
    2.09 ± 0.01 times faster than 'dash kakrun'

(bboxsh/links/sh is an Alpine 1.33 static version without builtin coreutils; all the \
others are 1_33_stable, STATIC + FEATURE_SH_STANDALONE + FEATURE_PREFER_APPLETS).

So busybox can also be about speed. Of course, this is well known, since ash has been \
much faster than bash for a long time, even without builtin applets (which is why \
distros switched to it at some point). \
_______________________________________________ busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox


[prev in list] [next in list] [prev in thread] [next in thread]