GNU bug report logs - #69872
29.2; Async native compilation of seq.el test uses up resources and hangs

Previous Next

Full log

Message #8 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Andrea Corallo <acorallo <at> gnu.org>
To: Jon Levin via "Bug reports for GNU Emacs, the Swiss army knife of text
 editors" <bug-gnu-emacs <at> gnu.org>
Cc: 69872 <at> debbugs.gnu.org, Jon Levin <hi <at> mhcat.ca>
Subject: Re: bug#69872: 29.2; Async native compilation of seq.el test uses
 up resources and hangs
Date: Mon, 18 Mar 2024 12:51:07 -0400

Jon Levin via "Bug reports for GNU Emacs, the Swiss army knife of text
editors" <bug-gnu-emacs <at> gnu.org> writes:

> When I update my installed packages on some version of emacs which
> is build with native-compilation enabled, most times I end up
> having to kill emacs and restart it in order to sidestep a problem
> where a single emacs process, seemingly running a compilation step
> (I think) uses 100% of a single virtual core and never completes.
>
> This is the process responsible:
>
>   mhcat      12800    6874 99 09:10 pts/96   00:10:41 /usr/bin/emacs -no-comp-spawn -Q --batch --eval (setq w32-disable-abort-dialog t) -l /tmp/emacs-async-comp-seq-tests-znp6r0.el
>
> I am not sure how to go about reproducing this problem from a
> standing start (emacs -Q) because this seems to be part of an
> otherwise opaque series of steps, and I'm not sure how it starts.
>
> Perhaps the author of seq.el can help me put together such a test.
>
> This is not a new problem, but I noticed that there is a plan to
> make native-comp part of the default emacs build, to which I have
> no objection, but feel that my experience is a data point which
> should not be overlooked. I usually end up building a
> non-native-comp version of emacs as a daily driver (I am lucky
> enough to have systems powerful enough that I notice no difference
> in performance with or without native-compilation enabled - aside
> from the initial compilation phases which in the case of
> native-comp, which are a minor inconvenience).
>
> Thanks!

Hi Jon,

I can reproduce on master with:

./src/emacs -batch -l comp -f batch-native-compile test/lisp/emacs-lisp/seq-tests.el

I'll try to have look.

I'm wondering as well why seq-tests is loaded in your system as well
being this a test file.

The change of performance of bytecode vs native really depends on the
workload so is no surprise that some of us experience noticable perf
uplifts and some less.

Bests

  Andrea

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.