llvm dev - Jul 2012 - [LLVMdev] Running c++-analyzer on svn r159506 llvm-clang

Hi,


Just for fun, I decided to run the svn r159506 llvm/clang/c++-analyzer
on svn r159506 llvm-clang. I just thought that it might be interesting
to run the analyzer on the llvm-clang codebase itself. I guess that
most people on this list will probably have at least some knowledge of
the llvm codebase, and be able to determine any false positives
relatively easily ?. And for the truly interested, maybe they can post
'false positives' against the Clang Static Analyzer component in the
llvm-clang bug tracker: http://llvm.org/bugs/

Anyway, the full report is located here:
http://lbalbalba.x90x.net/clang-analyzer/llvm-clang/



Regards,


John Smith

John, 

Thanks for making the results of the analyzer available.

Ted has recently added IPA support for C++, so the number of bugs reported
increased dramatically. In some cases our diagnostic clarity is still lagging
behind for C++ bugs that span multiple functions (it's WIP).

Cheers,
Anna.

On Jul 3, 2012, at 10:06 AM, John Smith wrote:
> Hi,
> 
> 
> Just for fun, I decided to run the svn r159506 llvm/clang/c++-analyzer
> on svn r159506 llvm-clang. I just thought that it might be interesting
> to run the analyzer on the llvm-clang codebase itself. I guess that
> most people on this list will probably have at least some knowledge of
> the llvm codebase, and be able to determine any false positives
> relatively easily ?. And for the truly interested, maybe they can post
> 'false positives' against the Clang Static Analyzer component in
the
> llvm-clang bug tracker: http://llvm.org/bugs/
> 
> Anyway, the full report is located here:
> http://lbalbalba.x90x.net/clang-analyzer/llvm-clang/
> 
> 
> 
> Regards,
> 
> 
> John Smith
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

On Tue, Jul 3, 2012 at 7:51 PM, Anna Zaks <ganna at apple.com>
wrote:
>
> Thanks for making the results of the analyzer available.
>
No problem. Im still just thinking about how I could assist in
improving both the compiler and the src analyzer. Running the analyzer
on the compiler seemed to be a sensible approach, as people here will
probably know (at least a little) about the both of them, which may
increase the small likelyhood that they either fix the compiler or the
analyzer.

>
> Ted has recently added IPA support for C++, so the number of bugs reported
increased dramatically.
> In some cases our diagnostic clarity is still lagging behind for C++ bugs
that span multiple functions (it's WIP).
>
Im aware that some of the program is still work in progress; the C++
analyzer does seem to be in a better state though than what the
current web pages imply. :)


If doing this is indeed helpful, and if it's not part of the build-bot
already, I have no problems with running the analyzer on a
semi-regular basis.


Thanks,


John Smith.