Massimiliano Culpo
Massimiliano Culpo
@tgamblin I don't see differences with `develop` on a test run on Marconi, so I think this is safe to merge. In case I am wrong, and we get user...
My understanding from last meeting is that #32537 is not needed to have pipelines rebuild all specs. It's an orthogonal feature that ensures each job in the Gitlab pipeline will...
> No, we do need that first, unless we change our approach to "spackbot rebuild everything". What is in that PR that is needed for rebuilding everything?
> A single spack install command That's the part I can't currently figure out from code. Why do we need a single command?
The test failing is: ``` lib/spack/spack/test/llnl/util/lock.py::test_write_lock_timeout_on_write_ranges_2[/run/user/1001] ``` which might be spurious. I'll trigger a re-run.
I was able to reproduce with: ```console % for ii in $(seq 10); do SPACK_PYTHON=python2 SPACK_TEST_SOLVER=clingo spack unit-test -k "installed_externals_are_reused" ; done ``` I can give it a try to...
As a quick mitigation we can mark the test at xfailing when Python 2.X is used
I added a print to the assert, and: ``` AssertionError: ext1: [email protected]%[email protected]+fee+foo arch=test-debian6-core2, ext2: [email protected]%[email protected]+fee+foo arch=test-debian6-core2 ``` so it seems to be an error in the fixture, i.e. both have...
Ping @OliverPerks
@eugeneswalker While trying to experiment with requirements, I made trilinos fail due to a conflict with variants and the version of CUDA used. Is this a good info, or was...