Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Looks like this test, when it was introduced/substantially refactored in D9426 - what was missing compared to all the other tests updated to run on linux, was the Makefile change to use libc++, so it was failing in places where libc++ wasn't the default. Fix that and it seems to pass for me, at least. Not sure if we need to keep some of the other expected failures (on some particular subset of clang, etc?)
Also there's a few other of these pretty printer tests that are still classified as skip-on-gcc, though at least some have been reclassified as passing-on-gcc in changes since D9426.
LGTM, thanks!
Tests with libc++ as a category are not run on Windows or when GCC is the test compiler, so those decorators are redundant. Removing the Clang XFAIL also seems fine, I would just see if any matrix bot complains and then re-apply it with a proper minimum required version.
It's a single bot that runs the test suite against a variety of compilers (well, only various versions of clang, I think)...
There is one on Green Dragon with a few Clang versions and I there is my Linux one with most GCC/Clang versions