Closed tshort closed 1 year ago
Sounds reasonable to me.. I wonder what the CI failures are
Yeah. Lots of CI failures. Nightly looks completely broken. Some of the macOS tests are breaking, too. Not sure what to do. Upgrading GPUCompiler may be enough.
Oh hmm, looks like a lot of these are LoopVectorization (or, relatedly, StrideArrayCore)-related. In particular the CI / Julia 1 - ubuntu-latest - x86
failure (which is a bit odd since CI / Julia 1.9 - ubuntu-latest - x64
passed fine) looks to be just failure to precompile LoopVectorization for some reason...
It might be worth re-running these in a day or two to see if any resolve on that end..
Some uses of
repr
end up creating long names with the modules prepended. If they are created in a temporary module, it's even worse. So, this PR substitutesnameof
instead to give just the name of the function.