Oliver Stöneberg

Spoiler alert: I start to ramble in the later part of this comment because those realization just started coming in and I didn't feel like re-doing the whole post. Sorry...

It could also help to expose gaps in the unit tests. The issue it found in the changes I did was not caught by those but unfortunately I was not...

Analyzing the code from https://trac.cppcheck.net/ticket/12861: Clang 20 - `11,437,539,974` -> `3,423,215,226` -> `737,254,030` This needs more testing as it might be detrimental in cases which are not affected by excessive...

> Can we reduce the timeout in the #12847 test with this? Yes. I forgot to add the reference for the code I tested it with.

> Clang 20 - `11,437,539,974` -> `3,423,215,226` -> `737,254,030` The second drop does not occur. It was a mistake in an intermediate version.

> > Clang 20 - `11,437,539,974` -> `3,423,215,226` -> `737,254,030` > > The second drop does not occur. It was a mistake in an intermediate version. There's a way to...

We do not detect this ourselves because of the conditional code inside. I wonder if we could add a flag which does not omit such warnings similar to `--emit-duplicates` but...

> We do not detect this ourselves because of the conditional code inside. This is just an assumption. I do not have actually checked this (yet).

> This is just an assumption. I do not have actually checked this (yet). I confirmed that it is indeed omitted because of the preprocessor checks. But I do not...

> But I do not see where the bailout happens in the code It is the `mTokenizer->hasIfdef()` call in `CheckClass::checkConstFunc()`.