I believe these locales were chosen simply because they can be considered
fairly common locales, but I'm only guessing.
IMO, the right way to solver fragmentation between LIT and testit is to remove
testit from the tree. It doesn't work on Linux out of the box, but can be made
to work with some modification. While that is easily fixed, it's non-parallel,
doesn't have very helpful output, and doesn't handle XFAIL in the first place,
let alone REQUIRES.
Adding documentation for requirements to the website would be great.
On a somewhat related note, I'm really hoping to finally look at zorg one of
these days so we can have buildbots set up (for Linux, FreeBSD, and Mac, at the
very least). That way we can do away with the manually generated and uploaded
test results and just link to the latest buildbot results.
================
Comment at: test/lit.cfg:192-193
@@ +191,4 @@
+ config.available_features.add('locale.{}'.format(loc))
+ except:
+ pass
+
----------------
Eric Fiselier wrote:
> Should we emit a warning here about the missing locale?
That's a good idea. LIT will mention that the tests are unsupported, but it
isn't explicit about why.
http://reviews.llvm.org/D4759
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
