Required for asterisk on arm, arm64, ppc and ppc64 with CC=clang. I have no way to test any of these. I'll attach a simple .c file that can be used to test the use case for asterisk at the very least. Reproducible: Always
Created attachment 682642 [details] test_clang_asterisk.c clang -fblocks -lBlocksRuntime -o test_clang_asterisk test_clang_asterisk.c $ clang -fblocks -lBlocksRuntime -o test_clang_asterisk test_clang_asterisk.c $ ./test_clang_asterisk Result: Success $ exit code will be 0 in case of success, or 1 in case of failure. It's probably best to merge BlocksRuntime itself with clang, but I just merged it with gcc ironically for the above test. Not seeing clang itself as keyworded for ppc, so ppc here might be moot unless clang gets keyworded for ppc as well. We can always retain the package.use.mask for asterisk on ppc to prevent asterisk requesting CC=clang in case of USE=blocks.
arm done
Looking good on ppc. # cat blocksruntime-765199.report USE tests started on Di 19. Jan 18:41:02 CET 2021 FEATURES=' test' USE='' succeeded for =sys-libs/blocksruntime-0_pre20171027-r1 USE='-static-libs' succeeded for =sys-libs/blocksruntime-0_pre20171027-r1 USE='static-libs' succeeded for =sys-libs/blocksruntime-0_pre20171027-r1 # clang -fblocks -lBlocksRuntime -o test_clang_asterisk test_clang_asterisk.c # ./test_clang_asterisk Result: Success
arm64 done
ppc done
(In reply to ernsteiswuerfel from comment #3) > Looking good on ppc. > Thank you!
(In reply to Jaco Kroon from comment #1) > Created attachment 682642 [details] > test_clang_asterisk.c > > clang -fblocks -lBlocksRuntime -o test_clang_asterisk test_clang_asterisk.c > > $ clang -fblocks -lBlocksRuntime -o test_clang_asterisk > test_clang_asterisk.c > $ ./test_clang_asterisk > Result: Success > $ > > exit code will be 0 in case of success, or 1 in case of failure. > > It's probably best to merge BlocksRuntime itself with clang, but I just > merged it with gcc ironically for the above test. > > Not seeing clang itself as keyworded for ppc, so ppc here might be moot > unless clang gets keyworded for ppc as well. We can always retain the > package.use.mask for asterisk on ppc to prevent asterisk requesting CC=clang > in case of USE=blocks. This could be added to src_test if it’s valuable.
(In reply to Sam James from comment #7) > (In reply to Jaco Kroon from comment #1) > > Created attachment 682642 [details] > > test_clang_asterisk.c > > > > clang -fblocks -lBlocksRuntime -o test_clang_asterisk test_clang_asterisk.c > > > > $ clang -fblocks -lBlocksRuntime -o test_clang_asterisk > > test_clang_asterisk.c > > $ ./test_clang_asterisk > > Result: Success > > $ > > > > exit code will be 0 in case of success, or 1 in case of failure. > > > > It's probably best to merge BlocksRuntime itself with clang, but I just > > merged it with gcc ironically for the above test. > > > > Not seeing clang itself as keyworded for ppc, so ppc here might be moot > > unless clang gets keyworded for ppc as well. We can always retain the > > package.use.mask for asterisk on ppc to prevent asterisk requesting CC=clang > > in case of USE=blocks. > > This could be added to src_test if it’s valuable. Sure, happy to "donate" the code. I'm just not sure how valuable and the extent of the coverage of the test. It's really very, very basic test aimed 100% at asterisk's requirements. Happy to add a copyright header etc similar to ebuilds indicating that it's provided "as is" and in the "public domain" (there really isn't anything copyrightable in that source that I can think of). Just about the only argument in favour of adding this IMHO is "any test is better than no test". Sergei, if you're interested in this I can do a PR, or you can add it, as you see fit.
I'd prefer tests to be added upstream.
(In reply to Sergei Trofimovich from comment #9) > I'd prefer tests to be added upstream. Thank you. We'll go with your opinion on this matter I reckon. So it's just PPC64 that's still outstanding?
ppc64 done all arches done
(In reply to Sergei Trofimovich from comment #9) > I'd prefer tests to be added upstream. I mean, sure, but they don't exist upstream, and this is a minimal way to at least see if it kind of works. Better than nothing.
