search

conda-forge / openblas-feedstock

A conda-smithy repository for openblas.
BSD 3-Clause "New" or "Revised" License
9 stars 38 forks source link

OSX: Drop build-prefix rpath #110

Closed jschueller closed 3 years ago

jschueller commented 3 years ago

The rpaths contain a BUILD_PREFIX entry, delete it to not confuse tools such as delocate

@conda-forge-admin please rerender

conda-forge-linter commented 3 years ago

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

github-actions[bot] commented 3 years ago

Hi! This is the friendly automated conda-forge-webservice. I tried to rerender for you, but it looks like there was nothing to do.

jschueller commented 3 years ago

@conda-forge-admin please rerender

jschueller commented 3 years ago

some linux jobs fail due to connection issues,

the real problem is some test crashes for the osx "ilp64" variant:

2020-11-18T10:04:32.3203600Z Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
2020-11-18T10:04:32.3203970Z 
2020-11-18T10:04:32.3204210Z Backtrace for this error:
2020-11-18T10:04:32.3244800Z #0  0x1066779cc
2020-11-18T10:04:32.3246980Z #1  0x106676dc5
2020-11-18T10:04:32.3251930Z #2  0x7fff6b0cb5fc
2020-11-18T10:04:32.3257810Z /bin/sh: line 1: 35444 Segmentation fault: 11  OMP_NUM_THREADS=2 ./xscblat2 < sin2
2020-11-18T10:04:32.3261810Z make[1]: *** [Makefile:112: all2] Error 139
2020-11-18T10:04:32.3263310Z make[1]: *** Waiting for unfinished jobs....
2020-11-18T10:04:32.6343910Z 
2020-11-18T10:04:32.6350710Z Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
2020-11-18T10:04:32.6351670Z 
2020-11-18T10:04:32.6352400Z Backtrace for this error:
2020-11-18T10:04:32.6355370Z #0  0x106fd59cc
2020-11-18T10:04:32.6356120Z #1  0x106fd4dc5
2020-11-18T10:04:32.6356860Z #2  0x7fff6b0cb5fc
isuruf commented 3 years ago

This should probably be a change in conda-build at https://github.com/conda/conda-build/blob/3912c85c787741c11d713038f5a62fc52708032b/conda_build/post.py#L399

jschueller commented 3 years ago

So now we're removing the host prefix instead of the build prefix, at first thought I expected the build_prefix to be removed.

-            macho.delete_rpath(path, rpath, build_prefix=prefix, verbose=True)
+            macho.delete_rpath(path, join(host_prefix, rpath), build_prefix=prefix, verbose=True)
isuruf commented 3 years ago

Yes, conda-build never removed the paths in build_prefix. Absolute paths in host_prefix are removed because relative paths are added.

jschueller commented 3 years ago

ok, so I could remove build_prefix rpath there, thanks!