On Mon, Oct 9, 2023 at 4:03 PM Patrick Robb wrote: > >> Hello, > > Yes, backporting would be ideal from a CI perspective because without it > we can't run arm64 testing on LTS tests. But I know there are other > considerations which also have to be weighed. > > David also has a patch[1] which should resolve the underlying issue which > introduces the failures on the unit test we want to skip. If that patch is > accepted, and backported, fixing our original problem with unit testing on > our arm testbeds, that's another solution, at least for this specific unit > test issue. > > It would still be nice to have this feature in case we need it otherwise. > > [1] > https://patches.dpdk.org/project/dpdk/patch/20230821085806.3062613-4-david.marchand@redhat.com/ > > Hi. just to close the loops on this, yes David's aforementioned patch did resolve the unit test failure which was preventing us from running fast-tests on our arm64 test beds. But, it is not (yet, at least) backported for LTS releases. Even if it were, having Bruce's patch here backported would mean the CI testing approach could be common across releases in situations where testcases have to be skipped. Anyways, whether it's possible or "worth it" is ultimately down to the community's bandwidth, but I didn't want to let the conversation lapse without an update, and raising what the benefits would be. In any case, thanks again Bruce for the rework, it's a great addition.