[cfe-commits] [patch] installing clang-check as part of clang

[David Blaikie]

On Thu, Jul 26, 2012 at 12:26 PM, Matthieu Monrocq
<matthieu.monrocq at gmail.com> wrote:
>
>
> On Thu, Jul 26, 2012 at 7:28 PM, Jordan Rose <jordan_rose at apple.com> wrote:
>>
>>
>> On Jul 26, 2012, at 10:23 , David Blaikie <dblaikie at gmail.com> wrote:
>>
>> > While reading the "How To Setup Clang Tooling for LLVM" documentation
>> > ( http://clang.llvm.org/docs/HowToSetupToolingForLLVM.html ) I ran
>> > into a snag where the document implied that clang-check would be
>> > installed alongside clang. This is currently not the case - we don't
>> > install clang-check, at least not in the cmake build (&, given the
>> > presence of "NO_INSTALL = 1" in the Makefile, I assume we don't in the
>> > make build either).
>> >
>> > Should we? It seems like a natural enough thing to install, though I
>> > realize the specifics of which tools will be developed where and how
>> > they'll be installed is still in flux, so I figured I'd start a thread
>> > to discuss this rather than just committing it.
>> >
>> > [as a side note: why do we install diagtool (perhaps there's some use
>> > for it other than the internal diagnostic flag regression testing?)
>> > and c-index-test (by name I would've thought that was just an internal
>> > test binary)]
>>
>> "diagtool tree" and "diagtool show-enabled" are at least theoretically
>> useful externally, though we don't document them anywhere. I kind of figured
>> things in tools/ are potentially useful outside of clang and could be
>> installed, and things in utils/ were just for internal use. I don't think
>> that was ever formally established, though.
>>
>>
> Is not diagtool used by the test Ted introduced to make sure clang does not
> regress in its list of warnings without flags ?

It is - though that doesn't explain why it's part of clang's
installation for end users of the compiler. (we have other internal
tools that aren't "installed" - I'm sort of surprised scan-build isn't
installed, for example - perhaps it's not considered production
worthy/relevant)

>
> -- Matthieu