Re: notmuch breaks on \. in config file with upgrade from glib2 2.76.1 to 2.76.5

[David Bremner <david@tethera.net>]

Eric Blake <eblake@redhat.com> writes:
> $ tail -n3 .notmuch-config
> [query]
> q1=from:/.*\\.example\\.org/
> q2=from:"/.*\.example\.org/"

> glib2-2.76.5-1.fc38.x86_64
> glib2-2.76.5-1.fc38.i686
> $ notmuch --config=$PWD/.notmuch-config config list
> $ echo $?
> 0
> $ notmuch --config=$PWD/.notmuch-config count
> $ echo $?
> 1

I could not duplicate the glib issue with glib 2.77.2 on Debian.
Looking at the source, it looks like we don't have 71b7efd08a1fe
(or equivalent).

╰─ (git)-[release]-% tail -3 bad-config
[query]
q1=from:/.*\\.example\\.org/
q2=from:"/.*\.example\.org/"
╭─ motzkin:upstream/notmuch/test/tmp.T030-config
╰─ (git)-[release]-% ../../notmuch --config=./bad-config count
52
╭─ motzkin:upstream/notmuch/test/tmp.T030-config
╰─ (git)-[release]-% echo $?
0

I usually like to start with a failing test, but it seems that may not
be possible here, since the actual failure only happens with specific
versions of glib.

From the discussion surrounding the revert / fix, it seems like the
behaviour of setting the GError, but also returning a string is itself
an error and can be expected to change. The assumption in the notmuch
code [1] is also that the status string only needs to be checked
on non-success status code. It should be OK to check the string anyway, but I
guess there might be some bugs uncovered.



[1]: notmuch.c, around line 547\r