[llvm] r334543 - Fix how LLVMOPTIONALCOMPONENTS is passed to llvm-build

[Ahmed Bougacha via llvm-commits]

On Wed, Jun 13, 2018 at 2:37 PM, Reid Kleckner <rnk at google.com> wrote:
> On Wed, Jun 13, 2018 at 2:33 PM Ahmed Bougacha <ahmed.bougacha at gmail.com>
> wrote:
>>
>> On Wed, Jun 13, 2018 at 11:56 AM, Reid Kleckner <rnk at google.com> wrote:
>> > Thanks for the revert. I agree, it did look quoted correctly, but I
>> > figured
>> > this was a workaround for some platform with busted quoting rules.
>>
>> No worries;  the author replied elsewhere: it turned out to be some
>> python configuration issue.
>>
>> > I don't see why passing an empty argument enables building the intel jit
>> > events library, but I have no interest in investigating it.
>>
>> " " evaluates as true in cmake
>
>
> Yes, but the llvm-build tool is python, and we are not checking
> LLVMOPTIONALCOMPONENTS anywhere for truth, so far as I can see.

Ah, sorry:  I mean, we set LLVM_USE_INTEL_JITEVENTS in cmake, and
that's used even after the llvm-build invocation.  So, it's not
enabled in llvm-build, but the rest of cmake code see it as enabled
(such as, in tools/CMakeLists.txt, the if() that predicates building
tools/llvm-jitlistener)

-Ahmed