Combining a type-inference tool for C with cppcheck

[Leandro Melo (ltcmelo)]

Hi everyone,

if I understand correctly, cppcheck's analyses are mostly local. Still, how complete is cppcheck's parsing and how important are the headers to it? I'm wondering if there are cases in which cppcheck either produces false positives or misses an issue particularly due to C's ambiguous syntax and absent type/var declarations.

I'm working on a type inference for C and I'm currently investigating how useful could it be when combined to static analysis tools such as cppcheck. (1) Suppose a few headers are unavailable (e.g. cross-platform dev), would inferring the absent declarations improve cppcheck's results? (2) Is there any reason why you would like to analyse a particular function (or set of functions) in isolation, without #including so many headers but still having all types declared is necessary to cppcheck?

I'd grateful if you can share any "story" in this front. I'm trying to assemble interesting use cases to be used in a paper. In the case you'd like to try-out the type-inference, there's an online interface available: http://cuda.dcc.ufmg.br/psyche-c/

Thank you for any feedback,
Leandro

[Daniel Marjamäki]

sorry this got stuck in some moderation ... is this thread still interesting?