On 8/24/25 6:39 AM, Mark Wielaard wrote:
Hi,
On Wed, Aug 13, 2025 at 08:53:44AM +0200, Richard Biener via Gcc wrote:
On Tue, Aug 12, 2025 at 10:43 PM Andrew Pinski via Gcc <gcc@gcc.gnu.org> wrote:
I think we should have a requirement of the bare minimum for a port is a
maintainer.
I also vote to have a testresults for the target at least once a year.
Having testresults also shows the port builds (as a cross at least) and is
able to build its target libraries. IMO broken ports are worse than
unmaintained ones, esp. if we release with those.
Ideally we'd have build bots with publically visible results that
test the building part (doesn't have to run often), this should
include building cross-binutils, newlib/glibc/avr-libc as fit.
Before releasing I'd auto-deprecate all broken configs, I'd expect
maintainers to analyze such failures (they might be bot issues).
Having testresults would be secondary (but nice), that includes
running the runtimes testsuite.
That said, I'd like to move away from gcc-testresults as a vetting
tool to something more modern.
If the port has a (qemu) emulator we could create a x86_64 container
for it and run it once a month/week on one of the faster
builder.sourceware.org workers.
That's what my tester does for alpha, m68k, hppa, sh4, ppc, s390.
Bootstrap, regression test, build glibc & kernel.
Jeff
