Hi Jordi,
Thanks for looking into that, and offering builds.
Le jeudi 25 septembre 2025 à 11:33, Jordi Mallach a écrit :
El dc. 24 de 09 de 2025 a les 14:22 +0200, en/na Benjamin Cama va
escriure:
Hi,
On Tue, 23 Sep 2025 16:55:22 +0200 Benjamin Cama <[email protected]>
wrote:
> Which makes me conclude that "sieve" is simply unusable because of
> this bug.
To clarify a bit: after using a patched version, this seems to be
because the destination of "fileinto" does not exist yet that I get
this
crash; after creating it (the patched version does it automatically),
it
works well, and maybe the current version, but I did not check.
OK so I checked trixie’s version in a chroot, it segfaults too.
But when digging a bit deeper on the test case, it is: having a filter
using fileinto in a directory that exists (I just mkdir’ed it), but has
nothing in it. So I think it may *look like* a maildir, but has not the
full structure.
I checked that sieve works nicely when the directory simply does not
exist, or when a maildir exists. So in the end this may be a very
particular use case… (or rather, user error)
Anyway:
I have built MU with the NULL pointer fix and another additional
segfault fix. I'd be grateful if you could test it before I upload to
proposed updates.
https://people.debian.org/~jordi/mu-trixie
changes and dsc files are signed with my Debian key.
Thanks a lot. I tried it in the chroot, and it kind of “work” by not
segfaulting anymore: but it does *nothing* and signals no error, which
may not be ideal. Excerpt in verbose mode:
sieve: test.sv:14.1: IMPLICIT KEEP dans le message uid 302
sieve: test.sv:14.1: IMPLICIT KEEP dans le message uid 303
sieve: test.sv:14.1: IMPLICIT KEEP dans le message uid 305
sieve: test.sv:14.1: IMPLICIT KEEP dans le message uid 306
sieve: test.sv:13.2-25: FILEINTO dans le message uid 321: livraison en cours
dans ~/Empty
sieve: test.sv:13.2-25: FILEINTO dans le message uid 326: livraison en cours
dans ~/Empty
sieve: test.sv:13.2-25: FILEINTO dans le message uid 327: livraison en cours
dans ~/Empty
Return code is 0. But no messages are moved.
So I thing you can either close this bug report for being a so peculiar
use case, or upload your fix which will make sieve not segfaults but may
surprise a user with its result.
I am sorry for this meh conclusion. It’s just my first look into
mailutils, and I already hit a second segfault (trying to use imap4flags
in a require, which actually does not exist…) and this gives me a not
too good impression on mailutils. Sorry for that, but thanks a lot for
your time and help.
Regards.
-- Benjamin
