Hi, On 30-08-2025 02:33, Carlos Henrique Lima Melara wrote:
That's indeed a consequence of using skip-not-installable.So, shouldn't we pin the unstable as -1 and only pin the package that triggered the regression testing to 900, in order to always make sure nothing else come from unstable?
No, because then the fallback wouldn't work and we'd get much more problems. autopkgtest already has an option to be stricter, but that caused more issues than it solved when I used that for some time on the ci.d.n infrastructure.
I think I might be missing some implications of this behaviour, right?
No, I tried. It was deliberately decided that what we now have is the less bad option.
Oh, well, we all regret something... But you can always remove it in a future major release ;-)
True. I don't regret it enough yet to go through that effort, but maybe others will.
Paul
OpenPGP_signature.asc
Description: OpenPGP digital signature
