Hi, On 14/04/25 at 16:35 +0200, Chris Hofstaedtler wrote: > Hi, > > not the maintainer, but having two questions: > > On Mon, Apr 14, 2025 at 01:28:57PM +0200, Lucas Nussbaum wrote: > > Source: gnocchi > > Version: 4.6.5-1 > > > During a rebuild of all packages in testing (trixie), your package failed > > to build on i386. > > 1) Why is this severity: serious? gnocchi is an Arch: all package, > supposedly only built on amd64-buildds.
My logic here is: arch:all can be used on all architectures. It's not always easy, from a build failure, to understand if failing to build translates to failing to work. So when it was clear that the package is not expected to work on i386, I did not file bugs. When it was not so clear, and possibly a bug in the package, I filed the bugs as severity serious. I'm totally fine if the bugs are turned into "severity=wishlist, tag=wontfix, title=i386 is unsupported and will never be". (at least it will be documented) > > > File "/build/reproducible-path/gnocchi-4.6.5/gnocchi/utils.py", line > > > 120, in timestamp_to_datetime > > > return datetime.datetime.utcfromtimestamp( > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^ > > > v.astype(float) / 10e8).replace(tzinfo=iso8601.iso8601.UTC) > > > ^^^^^^^^^^^^^^^^^^^^^^^ > > > OverflowError: timestamp out of range for platform time_t > > 2) This points to a property the project explicitly requested for > the i386 port. It would seem wrong to me to expect individual > packages to apply workarounds. I think you mean "rejected" not "requested". > Whats your take? I don't know. Maybe we need something like "architecture-has-64-bit-time_t"? Lucas
