[cfe-commits] [PATCH] CFG command line arguments for C++ features

[Ted Kremenek]

Looks good to me, although eventually in a future patch we should just have the CFG's BuildOptions be stored in AnalysisContext instead of replicating all these flags.

On Sep 29, 2010, at 11:42 PM, Marcin Świderski wrote:

> Thanks for noticing this. I will fix this and commit if there are no other issues.
> 
> W dniu 30 września 2010 08:25 użytkownik Zhongxing Xu <xuzhongxing at gmail.com> napisał:
> Hi Marcin,
> 
> Overall the patch looks good. But one problem:
> 
> @@ -126,7 +130,8 @@
>                                                      idx::TranslationUnit *TU) {
>    AnalysisContext *&AC = Contexts[D];
>    if (!AC)
> -    AC = new AnalysisContext(D, TU, UseUnoptimizedCFG);
> +    AC = new AnalysisContext(D, TU, UseUnoptimizedCFG, AddImplicitDtors,
> +        AddInitializers);
> 
> Notice that the fourth parameter is bool addehedges = false. We could set it to false for now. In the future we would add an option for it, too. 
> 
> You could spot this bug by actually running clang with the patch on a test case:
> 
> $ clang -cc1 -analyze -cfg-view -cfg-add-implicit-dtors scope.cpp
> 
> And see no effect with the new option.
> 
> I did test the patch some other way and it appeared to work properly. Sometimes tests can be broken though...
> 
> And for each patch, you can run 'make test' to ensure it does not break regression tests.
> 
> Yes, I do it all the time.
> 
> 2010/9/30 Marcin Świderski <marcin.sfider at gmail.com>
> 
> Patch adds two new command line arguments:
> -cfg-add-implicit-dtors - sets CFG::BuildOptions::AddImplicitDtors for AnalysisCosumer to true
> -cfg-add-initializers - sets CFG::BuildOptions::AddInitializers for AnalysisCosumer to true
> 
> Please approve for commit.
> 
> Marcin
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20100930/19544baf/attachment.html>