[cfe-commits] r159483 - /cfe/trunk/test/CMakeLists.txt

[Chandler Carruth]

Hmm, I see why. Not sure how to fix.

The way we use lit is really annoying in this situation.

There are two ways to tell lit about a test suite:
1) You can pass it a directory at the root of the test suite on the
commandline.
2) You can pass it a parent directory of the test suite and it will recurse
through them

There are then three ways to tell it about a site config file for the test
suite:
1) It can find one in the root of the directory passed on the commandline.
2) It can find one in a subdirectory that overrides the root one.
3) It can find one via a user parameter and manually override the site
config in the non-site config.

We currently seem to be using *all three* of the latter systems, and both
of the former systems. All at the same time.

The redundancy means that 'fixing' this at any one level doesn't actually
fix it... I don't even know how to fix it at all levels. =/

I would really like to remove #2 from both of these. We already have a
*local* config override that happens in subdirectories, being able to just
arbitrarily nest testsuites seems to cause more harm than good. =/ It would
have the side-effect of being the last thing required to fix the warning
message you see.


On Mon, Jul 2, 2012 at 2:44 PM, Jordan Rose <jordan_rose at apple.com> wrote:

> …nope, still see it. It's just a warning, though, so it's not a priority.
>
>
> On Jul 2, 2012, at 14:37 , Chandler Carruth <chandlerc at gmail.com> wrote:
>
> Give r159594 a whirl, and let me know how that looks?
>
>
> On Mon, Jul 2, 2012 at 2:09 PM, Jordan Rose <jordan_rose at apple.com> wrote:
>
>>
>> On Jul 2, 2012, at 14:06 , Chandler Carruth <chandlerc at gmail.com> wrote:
>>
>> On Mon, Jul 2, 2012 at 2:03 PM, Jordan Rose <jordan_rose at apple.com>wrote:
>>
>>>
>>> On Jul 2, 2012, at 13:53 , Chandler Carruth <chandlerc at gmail.com> wrote:
>>>
>>> > It was obvious: r159582. I can't promise that's *enough* but it seems
>>> definitely a good initial step. Let me know if that helps.
>>>
>>> Yes, it does! The only thing left is "Test suite 'Clang-Unit" contained
>>> no tests". And indeed, ClangUnitTests is not a target in my
>>> standalone-build project, but it and its subtargets do show up in my
>>> combined project.
>>
>>
>> Ah, ok, I have a patch for that to. It's also pretty obvious/easy when
>> you describe it. Give me a moment. =]
>>
>> That said, the patch will remove the message. We can't build the Clang
>> unittests in a standalone build. =/ At least, I'm not particularly
>> interested in adding the various build steps necessary for that -- i'd
>> rather make the not-standalone build just as fast for IDE users.
>>
>>
>> …unfortunate, but fair enough. (I see that the add_unittest function does
>> enough that it's not something we want duplicated in Clang's
>> CMakeLists.txt.)
>>
>> _______________________________________________
>> cfe-commits mailing list
>> cfe-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>>
>>
>
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120702/058dc3b3/attachment.html>