Bug#1116841: pybind11-json: FTBFS with the nocheck build profile

[Santiago Vila]

On Tue, Dec 09, 2025 at 09:41:05PM +0100, Bill Allombert wrote:
> On Tue, Dec 09, 2025 at 09:05:48PM +0100, Santiago Vila wrote:
> > found 1116841 0.2.15-1
> > thanks
> > 
> > Hello.
> > 
> > For some reason I fail to see, the 0.2.15-1 release made on 2025-11 by
> > Bill Allombert has undone all the changes in uploads 0.2.13-5 (which I
> > did on 2025-10 and already propagated to testing) and 0.2.13-4 (which
> > Graham Inggs did in 2025-01 and it's the version in stable).
> > 
> > As a result, we have to fix this RC bug again (and probably restore
> > the missing history, plus putting all the changes in salsa properly).
> 
> Hello Santiago, I do not understand what happened, but I have all the files
> locally so I will fix it. I am really sorry for the trouble.

[ So, I'm a little bit confused. If you are a team member and can commit
  to salsa, why the last upload says "NMU requested by Gordon" instead
  of "Team upload requested by Gordon"? ]

Anyway, since you offer to fix it, here is my suggestion: for the
consistency of history I would do a complementary commit like this one
before handling 0.2.15-1:


From: Bill Allombert <ballombe at debian.org>
Date: Sat, 1 Nov 2025 10:58:51 +0100
Subject: [PATCH] Revert releases 0.2.13-4 and 0.2.13-5

This is a fake commit to reflect the fact that the upload following
0.2.13-5 was mistakenly based on version 0.2.13-3.


(Such commit would be the outcome of "git diff debian/0.2.13-5 debian/0.2.13-3"
in the current salsa repo).

After that, the difference between 0.2.13-3 and the real 0.2.15-1 is
very small and will not be confusing to see in the repository.

But I leave that to you.

Thanks.