Details
- Reviewers
dsanders
Diff Detail
Event Timeline
MultiSource/Benchmarks/Prolangs-C/archie-client/Makefile | ||
---|---|---|
6–8 | Shouldn't this disable the test if re_comp is unavailable? It should be defined/not-defined by the configure step. |
MultiSource/Benchmarks/Prolangs-C/archie-client/Makefile | ||
---|---|---|
6–8 | We can't rely on the configuration script for this at the moment (see http://lists.llvm.org/pipermail/llvm-dev/2015-October/091758.html). Even if the configure step was reliable, disabling the whole test, seems like an overkill to me. Given the transition to CMake, I suggest that we commit this and properly configure the test-suite for cross-compilation with CMake in the future. |
MultiSource/Benchmarks/Prolangs-C/archie-client/Makefile | ||
---|---|---|
6–8 | I'm not keen to change this test in this way for all Mips targets because of a MUSL specific problem. I don't mind a temporary solution but I'd like it to target MUSL specifically. |
FWIW, I tested our out-of-tree buildbots and they are passing the test without any problem. This should be a temporary solution as long as we can't use rununder with the CMake build. With CMake it's really easy to check for the existence of these functions during configuration time.
Shouldn't this disable the test if re_comp is unavailable? It should be defined/not-defined by the configure step.