I have had another think and I would like to make the folloiwng points.
1.) license-reconcile is an experimental package and not at all critical
to Debian. Unlike decscripts so something is a really wrong if a bug in
license-reconcile affects devscripts.
2.) I think what Steve meant to do was clone #801609, retitle and
reassign the clone to devscripts. That's reasonable as there may well be
more work to do.
3.) As it is supposed to work, license-reconcile is not wholly dependent
on licensecheck. That said I think one filter is probably obsolete now
that licensecheck has been improved.
4.) The core thing I wish to avoid is licnese-reconcile development and
licensecheck development being tied together in anyway. That of course
means licensecheck will continue to cause problems for license-reconcile
but as long as those problems are not gratuitous I see no issue.
On 31/10/15 19:41, Dominique Dumont wrote:
On Friday 30 October 2015 10:58:55 Steve Langasek wrote:
But I'm also marking this as affects: devscripts, because I find it
surprising that the new licensecheck output includes a line for sample.png,
when the file was explicitly reported as unparseable. It doesn't seem
desirable to me that licensecheck would list files in its output that are
definitely not going to have embedded license/copyright information and
whose copyright information must be listed elsewhere.
Perhaps we want to make sure the new behavior for licensecheck is settled
before patching license-reconcile.
For what it's worth this change of behavior was requested in #794282 [1].
Jonas explicitly requested licensecheck to parse binary files.
This broke 'cme update dpkg-copyright', which led to #797562 [2] . To
alleviate the issue, I added a -t option to licensecheck to skip binary files.
After re-reading the whole saga, I think this situation can be improved. I'll
discuss with Jonas to nail down the requirements for a better solution.
All the best
[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=794282
[2] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=797562
