Davide Pesavento

I wonder what the gcovr implementation is doing here? Both mangled names are demangled to the same function name by `c++filt` (as expected), why isn't gcovr merging them based on...

The gcov json file (`indented-stream.cpp.2##0ecd6002d2762306549a1ce06cb21a16.gcov.json.gz`) already has the `demangled_name`, no? Why can't you use that? Are there cases where you do _not_ want to merge based on the demangled name?...

I checked the output of the Cobertura reporter and AFAICS constructors/destructors are properly merged now, and Jenkins doesn't complain about duplicate methods anymore. However it seems that they're being merged...

We're not using the JSON report so I'm totally fine either way (merging or not). I just wanted to double check that this was an intended behavior change.

Are you asking about the json report? I don't have a strong preference. I suppose adding the mangled name and keeping them separate is the more "lossless" approach, and also...

> Since there were no updates in 3 months: what's the fix in the meantime? Would downgrading to some non-latest version help? We've pinned our CI to commit 99b82e7 which...

> As this is your first pull request, [please be aware of the contributing guidelines](https://github.com/PCSX2/pcsx2/blob/master/.github/CONTRIBUTING.md). FYI the link to the contributing guidelines is broken.

Using commit f8168d652fb8cc78214c6cec02729ed051b40ac9 I'm getting an AttributeError. ``` (INFO) Attention, the decision analysis is experimental. It uses a fragile heuristic and depends on the code format. (INFO) Reading coverage data......

Thanks. I'm no longer seeing any errors with commit 2d19398215cf875bf85efe4a7e8b4af1ddcca4f5.

That's gone as well. Gcovr completes successfully now.