Update HierarchicalMNS for GCCcore toolchain#2870
Update HierarchicalMNS for GCCcore toolchain#2870boegel merged 4 commits intoeasybuilders:developfrom
Conversation
|
Ok, I reproduced the error that was seen for @Micket , this fixes it. |
|
This change can perhaps also make it possible to combine CUDA with GCCcore instead of GCC to make the GCC-CUDA based stuff.... might be worth looking at in the future... |
| tc_comp_name, tc_comp_ver = tc_comp_info | ||
| # Stick to name GCC for GCCcore | ||
| if tc_comp_name == GCCCORE: | ||
| tc_comp_name = 'GCC' |
There was a problem hiding this comment.
Are you sure this won't cause things built with GCCcore to end up in .../Compiler/GCC instead of Compiler/GCCcore
I'm not sure exactly what this affects.
There was a problem hiding this comment.
No, it won't affect that since GCCcore is built with the dummy compiler (so the condition non_dummy_tc 4 lines up is not satisfied). This is only relevant to compilers built with GCCcore and only affects the searching for matches in COMP_NAME_VERSION_TEMPLATES (see the for loop around this block).
I'm not overly confident that this has no impact though but I don't believe it does. As far as I know Clang is the only compiler build from source using GCCcore (rather than GCC).
There was a problem hiding this comment.
I just tried this out with Clang, and then with a normal package
eb Ghostscript-9.27-GCCcore-8.2.0.eb --include-module-naming-schemes hierarchical_mns.py
and it was placed correctly (under software/Compiler/GCCcore/Ghostscript )
|
Thanks for confirming the fix @Micket! |
@Micket With our custom naming scheme (related to HMNS) I was also seeing an issue with the Clang easyconfig in easybuilders/easybuild-easyconfigs#8254 This PR fixes that for me, however the reported error (for me) was:
This is quite different from what you reported in
easybuilders/easybuild-easyconfigs#8254 (comment)
Does this PR fix your problem?