[cfe-dev] C++ analysis vs C analysis

[Ted Kremenek]

By "internally", I mean those actively working on the analyzer.  There's no use getting a flood of bug reports from casual users for the 90% cases that would trigger just by the analyzer developers running the analyzer over a handful of C++ codebases.

On Feb 21, 2011, at 1:54 PM, Ted Kremenek wrote:

> It's not worth filing C++-specific bugs against the analyzer at this point.  We should internally vet it first by running it over a ton of code, catching crashes, etc.
> 
> On Feb 21, 2011, at 1:30 PM, Jean-Daniel Dupas wrote:
> 
>> I was waiting that the C++ analyzer be a little more mature, but if you tell it is worth filling bugs now, be sure I will do it.
>> 
>> Le 21 févr. 2011 à 20:11, Argyrios Kyrtzidis a écrit :
>> 
>>> It'd be a good idea to file bugs for the false positives when analyzing C++ code so we can keep track of them.
>>> 
>>> -Argiris
>>> 
>>> On Feb 21, 2011, at 8:14 AM, Jean Baptiste LE STANG wrote:
>>> 
>>>> OK, thanks. At least a result with false positive is better than no
>>>> result at all. I'm going to try with a recent SVN version.
>>>> 
>>>> On Mon, Feb 21, 2011 at 5:06 PM, Jean-Daniel Dupas
>>>> <devlists at shadowlab.org> wrote:
>>>>> 
>>>>> Le 21 févr. 2011 à 16:57, Jean Baptiste LE STANG a écrit :
>>>>> 
>>>>>> Hello,
>>>>>> 
>>>>>> I'm trying to use LLVM & scan-build (clang 2.8, checker-255) to
>>>>>> achieve a static analysis of  a C++ program to detect potential bugs.
>>>>>> Before doing it on my real program, I've been trying to make it work
>>>>>> on a simple program :
>>>>>> 
>>>>>> int main(int argc, char * argv[] , char * arge[]){
>>>>>>    int i;
>>>>>>    if (i<5){
>>>>>>            i = 50;
>>>>>>    }
>>>>>>    return i;
>>>>>> }
>>>>>> 
>>>>>> First case as a C program : scan-build -k -V -v gcc main.c
>>>>>> 
>>>>>> ANALYZE: main.c main
>>>>>> main.c:4:14: warning: The left operand of '<' is a garbage value
>>>>>>    if (i<5){
>>>>>> 
>>>>>> Second case as a C++ program :  scan-build -k -V -v g++ main.cpp and
>>>>>> i'm missing the previous error detected in scenario 1
>>>>> 
>>>>> The static analyzer does not support C++ yet.
>>>>> The svn version starts to support it but it is not ready to use AFAIK.
>>>>> I tried it last week, and it reported a lot of false positives.
>>>>> 
>>>>> -- Jean-Daniel
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>> 
>>>> _______________________________________________
>>>> cfe-dev mailing list
>>>> cfe-dev at cs.uiuc.edu
>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>>> 
>> 
>> -- Jean-Daniel
>> 
>> 
>> 
>> 
>> 
>> _______________________________________________
>> cfe-dev mailing list
>> cfe-dev at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>