Hi,

During a rebuild of all packages in sid, your package failed to build
on amd64.


Relevant part (hopefully):
The full build log is available from:
http://qa-logs.debian.net/2022/02/12/git-buildpackage_0.9.25_unstable.log

A list of current common problems and possible solutions is available at
http://wiki.debian.org/qa.debian.org/FTBFS . You're welcome to contribute!

If you reassign this bug to another package, please marking it as 'affects'-ing
this package. See https://www.debian.org/Bugs/server-control#affects

If you fail to reproduce this, please provide a build log and diff it with mine
so that we can identify if something relevant changed in the meantime.

control: reassign -1 pydoctor
control: force-merge 1005338 -1
control: affcts -1 git-buildpackage
thanks

Hi Lucas,

That's a problem in one of the build-deps as far as I can tell.

Cheers,
 -- Guido

Dear pydoctor maintainers,

I moved this here since nothing did change on the gbp site in that
area however it seems pydoctor didn't change either.

The python 3.8 changelog has

Python3 changelog and no sax parser change popped out that
python3's sax parser changes didn't

could that trigger it? I would expect for this to get noticed earlier
then though. I didn't look at changes in twisted yet.

Please move the issue back to gbp if you think it should be fixed there.

Cheers,
 -- Guido

Hi,

Am Sun, Feb 13, 2022 at 12:51:47PM +0100 schrieb Guido Günther:

I think the root for the issue is mainly the updated pydoctor package.
No other package that is a dependency of pydoctor has got a significant
update too and no other involved package got also an update since the
build of git-buildpackage is failing.

To dive into the problem I've prepared a local pbuilder chroot with the
data from the day before pydoctor 21.12.1-1 was uploaded (that was 15
Jan 2022 08:49:16 +0000) to unstable by using snapshots.d.o.

Building git-buildpackage with packages from 14 Jan 2022 works as
expected.

Using data from snapshot.d.o. right after the upload of pydoctor
21.12.1-1 makes the build of git-buildpackage fail with the exact same
error log visible in the starting email of this report.
I've added some hacky debug printing into
usr/lib/python3.9/xml/sax/expatreader.py around line 221 to see what
excately is the string the build is complaining about. This leeds to
this output (data is the second argument within 'def feed()' in line
206 of expatreader.py):

----%<----
+++debug+++
data=
--><div><tt class="rst-docutils literal">str</tt> or <tt class="rst-docutils literal">list</tt> of <tt class="rst-docutils literal">str</tt></div><--
undefined entity: line 1, column 46
+++debug+++
Traceback (most recent call last):
  File "/usr/lib/python3.9/xml/sax/expatreader.py", line 219, in feed
    self._parser.Parse(data, isFinal)
xml.parsers.expat.ExpatError: undefined entity: line 1, column 46

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/twisted/web/_flatten.py", line 390, in _flattenTree
    element = next(stack[-1])
---->%----

To me the data fiels looks like valid HTML/CSS code...

So I think that something inside pydoctor is now working different
between the new and the old version 19.11.0+git20200303.47424e7-1 in
the archive and the updated twisted package can't get blamed. twisted
was updated a few days later and then the build of git-buildpackge was
already broken.

Also the update of pydoctor was the first update for almost two years!
Given the changelog of pydoctor [1] is rather long since version 19.11.0
and I personally don't know enough how all parts for creating the
documantation of git-buildpackage is working together I've tagged this
report for help.

Maybe the issue can be reproduced without all the other working stuff
from git-buildpackage, if someone can help out how to get the relevant
code parts together it would be get easier to debug the stuff I guess.

[1] https://github.com/twisted/pydoctor/blob/master/README.rst

Regards
Carsten

control: retitle -1 Fails on non breaking space

Hi,

Carsten extracted this:

  <div><tt class="rst-docutils literal">str</tt> or <tt class="rst-docutils literal">list</tt> of <tt class="rst-docutils literal">str</tt></div>

and it would fail on   which came about by a non breaking space (C2
A0) in gbp's source code. Removing these makes gbp's API build pass
again but it still looks like a regression in pydoctor.

Cheers,
 -- Guido