From: Xiyue Deng <manphiz@gmail.com>
To: Eli Zaretskii <eliz@gnu.org>
Cc: 73926-done@debbugs.gnu.org
Subject: bug#73926: 29.4; *-autoloads.el generation not reproducible
Date: Mon, 21 Oct 2024 14:31:06 -0700 [thread overview]
Message-ID: <87msixuorp.fsf@debian-hx90.lan> (raw)
In-Reply-To: <86r089hkn7.fsf@gnu.org>
Hi Eli,
Eli Zaretskii <eliz@gnu.org> writes:
>> From: Xiyue Deng <manphiz@gmail.com>
>> Date: Mon, 21 Oct 2024 01:00:14 -0700
>>
>> I was trying to debug a reproducibility test failure on muse-el in
>> Debian Salsa CI reprotest[1]. It looks like that the contents of the
>> generated autoload file depends on the input sequence of files and
>> directories, and it looks like the `loaddefs-generate' doesn't enforce
>> such a sequence to be consistent.
>>
>> I have provided a patch (attached) to sort the input directories and
>> files in an attempt to fix this reproducibility issue.
>
> Did you time the modified code wrt the original one, especially when
> generating loaddefs for the entire Emacs tree? If the performance hit
> is significant, we'd probably want to make this behavior optional,
> like the other measures in Emacs that are meant to produce more
> reproducible builds.
>
> Also, AFAIK on MS-Windows directory-files produces a sorted list by
> default (just in the reverse order), so in that case calling 'sort' is
> not needed in the first place.
>
> I also wonder whether this will produce the same order on different
> operating systems, what with the file-encoding being a factor.
>
It looks like my previous test was actually incorrect and turned out
this patch on Emacs didn't actually fix the specific issue for muse-el.
So I'd like to drop this for now until I have a real fix for muse-el
first.
>> --- a/lisp/emacs-lisp/loaddefs-gen.el
>> +++ b/lisp/emacs-lisp/loaddefs-gen.el
>> @@ -596,9 +596,13 @@ instead of just updating them with the new/changed autoloads."
>> (concat "\\`[^=.].*" (regexp-opt tmp t) "\\'")))
>> (files (apply #'nconc
>> (mapcar (lambda (d)
>> - (directory-files (expand-file-name d)
>> - t files-re))
>> - (if (consp dir) dir (list dir)))))
>> + (sort
>> + (directory-files (expand-file-name d)
>> + t files-re)
>> + 'string<))
>
> Isn't string< the default when 'sort' is called to sort strings?
>
Yes, actually `directory-files' should have already sorted the results
using 'string-lessp as the documentation pointed out, so this sort is
not necessary. The one with the dir may worth considering, but will
test more before resubmitting.
> Thanks.
--
Regards,
Xiyue Deng
next prev parent reply other threads:[~2024-10-21 21:31 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-21 8:00 bug#73926: 29.4; *-autoloads.el generation not reproducible Xiyue Deng
2024-10-21 9:26 ` Eli Zaretskii
2024-10-21 21:31 ` Xiyue Deng [this message]
2024-10-22 5:06 ` Xiyue Deng
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87msixuorp.fsf@debian-hx90.lan \
--to=manphiz@gmail.com \
--cc=73926-done@debbugs.gnu.org \
--cc=eliz@gnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this external index
https://git.savannah.gnu.org/cgit/emacs.git
https://git.savannah.gnu.org/cgit/emacs/org-mode.git
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.