Am 07.01.2011, 10:27 Uhr, schrieb <[email protected]>:

> Hi,
>
> On Jan 4, 2011, at 13:58 , ext variadic.template wrote:
>
>> Hi,
>>
>> as i am working with the clang-libraries to parse objective-c-code,
>> and using qtcreator at the same time, im asking myself if there
>> are any (long term) plans to integrate clang into qtcreator or any
>> person currently tries to do so. I think this would be a big step
>> towards more refactoring/automation of some coding inside of
>> qtcreator + the ability to integrate code-analyzes. For sure, this
>> would have a big impact on the code of qtcreator and is nothing
>> that can be done quickly, also, it means another thirdparty-
>> dependence. But maybe there is something going on inside the labs..?
>
> There are five (possible) parts in the Qt Creator + clang story, which  
> are largely independent. Including status, these parts are:
>
> Compiling Qt with clang: the master branch in gitorious has an  
> experimental spec for clang on MacOS. It's in the unsupported dir. Both  
> we and the clang guys/girls regularly try to compile Qt with clang, so  
> work is ongoing in that area. Oh, and note that you will have to use a  
> very up-to-date version (meaning: newer than what Apple ships).
>
> Qt Creator tool-chain support for clang: no, not there. We will have to  
> support this when Apple starts using clang as the primary compiler on  
> MacOS. If you would like to give this a try, you might want to use  
> liblibclang.{so,dll,dylib} or libclangDriver.{a,lib} to give you the  
> right output instead of trying to figure out what the command-line tool  
> is spitting out. The clang website has quite some info about the driver.
>
> Clang as code-model: we investigated this a couple of times last year,  
> and since about December the most time-critical part (code completion)  
> is fast enough to use it. We still have to see if there are any legal or  
> distribution-technical hurdles. After that, then yes, we would like to  
> use clang here. However, it will take some time to move all currently  
> existing functionality from our current code-model to one driven by  
> clang. So don't expect it to happen in the next month ;-)
>
> Static analyser tool: we are working on infrastructure in Qt Creator to  
> run profilers/analysers/etc. The clang static analyser would be just  
> another tool to hook in there.
>
> Static analysis of Qt with clang: this has the same legal and  
> distribution implications as "clang as code-model". However, it is lower  
> on our wish-list. So: lots of ideas, currently no time, so won't happen  
> in the near-future if nobody is going to spend his spare time (or  
> Creative Friday) on it.
>
> Cheers,
> Erik.

Thanks Erik and Tobias for these informations. It's good to hear that
your doing some investigations on clang-integration. Im looking forward
to see a clang-based code-model and some clang-based analyser-tools inside
qtcreator (as it is the best IDE for linux, in my eyes) - not "next month",
but in the forseeable future :)
_______________________________________________
Qt-creator mailing list
[email protected]
http://lists.qt.nokia.com/mailman/listinfo/qt-creator

Reply via email to