[libcxx-commits] [PATCH] D84801: [libcxx] [test] XFAIL midpoint.float.pass on AArch64 with compiler-rt

[Sergej Jaskiewicz via Phabricator via libcxx-commits]

broadwaylamb added inline comments.


================
Comment at: libcxx/utils/libcxx/test/config.py:283
+
+        if self.get_lit_conf('use_compiler_rt'):
+            self.config.available_features.add('use_compiler_rt')
----------------
ldionne wrote:
> broadwaylamb wrote:
> > broadwaylamb wrote:
> > > ldionne wrote:
> > > > Not super happy about this for a couple of reasons:
> > > > 
> > > > 1. It adds more to `config.py`, which I'm trying to get rid of.
> > > > 2. There's nothing associated to the `use_compiler_rt` feature except the Lit feature itself. It's kind of like adding a new feature whose sole purpose is to `XFAIL` this specific test. Instead, it might make more sense to detect whether subnormal floats are supported?
> > > > 2. There's nothing associated to the `use_compiler_rt` feature except the Lit feature itself. It's kind of like adding a new feature whose sole purpose is to `XFAIL` this specific test. Instead, it might make more sense to detect whether subnormal floats are supported?
> > > 
> > > Do you mean like compile and run a test program that makes use of subnormal floats and generate a Lit feature based on whether it succeeds?
> > > 
> > > 
> > @ldionne could you please explain what you've meant?
> Yes, this is what I meant. I think it would provide more value because it would be detected automatically instead of being set explicitly by users. The test suite is quite complicated to configure today and I'd like to make it simpler. Detecting this automatically would, I think, help us go towards that. WDYT?
The idea was that when D85032 lands, the XFAILed test will unexpectedly pass, so we don't forget to update that test.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D84801/new/

https://reviews.llvm.org/D84801