(Adding @davidlt)

See also same change proposed for tang: https://src.fedoraproject.org/rpms/tang/pull-request/18#

Fine with me, but have you considered changing %meson_test on riscv64 to pass --timeout-multiplier 10 for all packages that use the macro? It would avoid having to potentially update tons of packages.

In any case, feel free to merge this and build, and thanks!

So far we have <300 modified packages in Fedora/RISCV dist-git overlay, and the smallest delta between downstream and upstream Fedora Kojis was ~1500 packages, where ~1000 just doesn't build in general (see mass rebuild results). Thus it's not that many packages that we modified timeouts (at least today), and hardware is getting faster.

Otherwise both approaches work for me. I will leave this up to @rjones and meson macro maintainers.

It definitely would be easier to switch it off at some point in one place globally.

I will examine what it would take to modify the macro & get back.

So turns out modifying the %meson_test macro where it is defined is very hard indeed. Adding just the extra parameter is possible, but making it conditional on %ifarch riscv64 seems impossible as far as I can tell. The rpm syntax used in macros files is somehow different from the regular syntax and I couldn't work out how.

The larger problem is that all of this stuff is upstream in the meson project (rather than in Fedora dist-git) requiring a patch there, which is... difficult.

Thanks for checking! Let's go with the patch here then.