[PATCH] Modeling APIs in the Static Analyzer

[Anna Zaks]

> On Jul 13, 2014, at 6:11 AM, Gábor Horváth <xazax.hun at gmail.com> wrote:
> 
> Hi Anna,
> 
> Thank you for the review. I have tweaked the test, so it no longer requires the error reporting tweak that is not done yet to pass. I have also added some high level comments to some files, if you think some information is lacking I will add them in the next iteration as well. The BugReporter patch is now separated into a different patch. 
> 
> 
> On 11 July 2014 18:02, Anna Zaks <ganna at apple.com> wrote:
> 
> For example, modeling functions should allow you to find bugs and suppress false positives outside of those functions. I would suggest adding a few of those tests first.
> 
> 
> How are the false positives suppressed? I did not find any resource on that. Found some analyzer attributes but I did not find them suitable for this purpuse at the first glance. But I think once the locations that are in a model file are omitted from the report path, the regular methods for suppressing false positives should work (and I will definitely add test case to ensure this once it is done).
> 

What I meant is that it is possible to construct a test where ability to model a function would eliminate a false positive. This would be another way to test your patch without worrying about BugReporter.

> Thanks,
> Gábor
> <api_modeling.patch><bugreporter.patch>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140714/4b31ae80/attachment.html>