On Mon, 04 Aug 2025 10:02:50 +0200 Jonas Smedegaard <[email protected]> wrote:
Quoting Matthias Geiger (2025-08-03 14:38:59)
> On Sun, 03 Aug 2025 09:53:00 +0200 Jonas Smedegaard <[email protected]> wrote:
> > I have tried to avoid benchmarking tests to get rid of the dependency
> > on criterion, but then another no-panic test mysteriously (to me) fails.
> >
> > I have pushed my changes to https://salsa.debian.org/debian/rust-ahash
> > and would appreciate if someone more clueful than me could try make
> > sense of why no-panic test fails.
> >
> IMO benchmark shouldn't be run anyway in Debian as they are a metric for
> upstream, but I disgress.
Yes, I agree.
> The reason they still run is that the [bench]
> section is still present in Cargo.toml, you need to actively disable it
> (i.e. patch it out
> [0]).
I admit that I this time chose an unusual more lightweight approach, but
believe to have succeeded in disabling them. With a more expansive patch
that removes the entries, and also removes the test files (because they
are places below tests/ not bench/ so gets autodetected as tests too),
I get the same error related to no-panic.
After some fiddling I was able to get ahash to build just fine with
sbuild, with all build time tests passing. From a fresh git clone I had
to disable patch 1003_version, since for some reason dpkg-source claimed
it was already applied. autopkgtest did not run since it's targeting
unstable and complains about a missing zerocopy. Any blockers to get
this uploaded to experimental ? Currently, this is blocking me from
updating glycin in experimental.
best,
werdahias
