RE: [cdt-patch] CDT2.0 Scanner configuration discovery

[Vladimir Hirsl <vhirsl@xxxxxxxxxx>]

> Vladimir,
> 
>   I have a chunk of code which rips through a project and extracts
> all of the headers from the source and then for each header if it
> is local, it attempts to determine if there is a local include path
> which should be created to satisfy it.
> 
>   I don't know if this fits in with your work, but I'm happy to 
> provide it as more of the "autoconfiguring the scanner" type of
> tools.
> 
> Thanks,
>  Thomas

Thomas,
thanks for offering your help, it is greatly appreciated.

Right now I am concerned with capturing scanner config from the make build 
output. The output should have the most accurate scanner config 
information, used by the compiler, so theoretically by providing the same 
scanner info to CDT parser it should produce the same results as the 
compiler would. 

There are couple of issues with this approach.
First, the CDT parser is still not 100% compatible with the GNU compiler.
Second, the user may not be able to build the project (or at least not 
successfully) so the captured scanner config may be incomplete.
Third, since the discovered scanner config is managed on a project level 
(for now), there may be conflicting entries, therefore scanner config may 
be incorrect.

In any of these cases we will need to apply other heuristics to add/fix 
entries in the scanner config. In that respect your suggestion is 
definitely worth looking into.

Thanks,
Vmir