Daniel Marjamäki

> I think slightly improving the fuzzing of garbage code is fine. Problem is that in my experience it's too easy to trigger crashes on garbage code. As far as...

I believe we can have a ready made script in the repo that I can run if I want to run this fuzzing locally. So feel free to consider creating...

> The exhaustive option should be switched on by default. Only if there are performance issues, users can decide to set it down to normal, while they need to be...

> And with the checkersReport message appearing on any analysis it seems that we are just throwing unnecessary useless information at the user (I don't even understand how this would...

> We could also include a mode which does no ValueFlow at all and call that shallow (or similar). Well I fear that such mode would have a drastic number...

> By reading the documentation I fear that most users does not read the documentation much and will not learn about this solution to their problem then. My idea with...

> And the current exhaustive is also a bit misleading as it might indicate that we are doing more than necessary - which isn't the case as it is the...

I ran --check-level=fast and --check-level=normal on 100 random packages. Total number of reports for each severity: error: 1634 my_check_diff_fast.log 1665 my_check_diff_normal.log warning: 5103 my_check_diff_fast.log 5112 my_check_diff_normal.log style: 15629 my_check_diff_fast.log 15873...

Refactorings will be needed before I merge this to main. I feel that ideally there should not be lots of copy/pasted code between valueflow.cpp and valueflowfast.cpp. I want to reuse...

ok `--check-level=fast` must be used explicitly by the user. It is not "fast" by default. Therefore in my opinion we should not write warnings that the analysis is fast and...