[PATCH] D20221: [test-suite] parameterization of llvm_{multi, single}source()

Artem Belevich via llvm-commits llvm-commits at lists.llvm.org
Tue May 31 16:24:33 PDT 2016


On Wed, May 18, 2016 at 6:59 PM, Matthias Braun <matze at braunis.de> wrote:
>
> > The problem is that it all happens under the hood and we, generally
> speaking, don't know the name of real target created by llvm_multisource()
> or llvm_singlesource() and thus can't use regular cmake API to augment
> rules to build that target.
>
> >
>
> > Perhaps that's the first thing I should fix. Once we have a way to
> figure out the name of the target we've generated, we can do custom things
> outside of SingleMultiSource.cmake. It will also make it possible to make
> llvm_singlesource to be just a wrapper over llvm_multisource with a single
> source file and thus avoid divergence you're concerned about.
>
>
> I completely agree here, I would describe it as: llvm_multisource() should
> not exist and people should just directly call test_suite_add_executable()
> instead. (llvm_singlesource() is already a wrapper over that).
>
> We really should have a way to know the target name. Maybe have a function
> that returns the target name when given a test name or maybe we can
> restrict the usage of llvm_target_prefix() to llvm_singlesource() so that
> at least you have a natural correspondence between test name and target
> name for all other cases...
>
>
I've sent D20842 to simplify executable name generation and provide a way
to return executable names to user.


-- 
--Artem Belevich
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160531/78545089/attachment.html>


More information about the llvm-commits mailing list