Bug#1001625: python3-script-but-no-python3-dep unoverridable
Hi Marc,
On Mon, Dec 13, 2021 at 4:33 AM Marc Haber
<mh+debian-packages@zugschlus.de> wrote:
>
> I wondering whether the missing leading backslash in the path is
> correct.
You probably meant a regular slash, and not a backslash. Lintian has
used such relative-looking path names since before it found me. The
only exception I know of is the hint annotation, which is usually
taken verbatim from your code. It can include leading slashes (for
conffiles, for example, which are absolute).
The path you asked about is part of a pointed hint. It includes file
information that will soon be a live link on our website. [1] Those
paths are taken straight from Lintian's in-memory copy of the file
index. [2]
[1] https://bugs.debian.org/743226
[2] https://salsa.debian.org/lintian/lintian/-/blob/master/lib/Lintian/Index.pm#L393-396
> debian/sudo-ldap.lintian-overrides:sudo-ldap: python3-script-but-no-python3-dep /usr/bin/python3 [usr/share/apport/package-hooks/*]
Our Git version already moved past your 2.114.0 with respect to
override treatments. We now allow globbing patterns [3] using
Text::Glob. [4] Unfortunately, the square bracket I recently selected
for file pointers is cumbersome to escape. The backslash \[ does not
work, while [[] (from bash) and the question mark ? both work. You
could leave out the ending ] or use the even more awkward-looking []].
None of that is great for users.
Overrides are currently in flux. Maybe overrides should be in a more
capable format, such as Deb822. In your case, that would probably look
something like:
Tag: python3-script-but-no-python3-dep
Note: /usr/bin/python3
Pointer: usr/share/apport/package-hooks/*
Perhaps I should also add a tool for users to generate or modify
overrides interactively. I have used something similar for years in
Lintian's test suite to recalibrate tests for sweeping changes.
Kind regards
Felix Lechner
[3] https://salsa.debian.org/lintian/lintian/-/commit/139009d5a54225ebff4509ec37b979cb898c17fe
[4] https://metacpan.org/pod/Text::Glob
Reply to: