[llvm-dev] Is ist a good idea to use lit and other test tools for non llvm projects?

[Dan Liew via llvm-dev]

On Thu, 24 Jan 2019 at 20:21, whitequark via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
>
> On 2019-01-24 20:17, David Greene via llvm-dev wrote:
> > alexp via llvm-dev <llvm-dev at lists.llvm.org> writes:
> >
> >> I have a project which uses llvm, but is (at least not yet) a
> >> contribution / not in the source tree of llvm.
> >> Is it a good idea (e.g. instead of using boost test framework) to use
> >> the llvm testsuite related tools in this case?
> >
> > That probably depends on the project, but I've long wished LLVM's
> > testing tools were available outside of LLVM, packaged up separately so
> > they could be more widely used.  They're great tools!
>
> lit is, in fact, packaged separately--it is available on PyPI.
> FileCheck isn't; there is a similar tool on PyPI, called OutputCheck,
> which I have used extensively for similar jobs (matching compiler IR)
> but
> I like it much less than FileCheck.

Ha. I'm actually the author of OutputCheck. I wrote it for the STP
constraint solver a few years ago and I made it because I wanted
something like `FileCheck` but without depending on LLVM. It's very
much a poor man's FileCheck. It only has a small subset of FileCheck's
functionality.

I had no idea people were actually using it.

I don't really work on the project anymore but I'm sure the STP
developers would be more than happy for some to work on it.

Thanks,
Dan.