Mauricio Collares

> Just add that and be happy. I see this was used as justification for the actual fix. Just checking, does this take care of possible overlaps inside the array?...

Ok, I think I see. `max long long` is not 63, but rather 53 because the code internally stores the numbers as doubles and probably enforces it all fits exactly...

https://github.com/WebAudio/web-audio-api/issues/2033

https://github.com/WebAudio/web-audio-api/issues/2032

You can have both packages, but the command to see goals in `lean4-mode` is `lean4-toggle-info` (bound to `C-c TAB`). The command you cited is from the `lean-mode` package (i.e., Lean...

@grunweg Future idea for this linter: Catch #7678 as well.

Here's further evidence that the particular versions are not the problem. Running the minimized testcase with R 4.3.3 and Python 3.12 under `valgrind --trace-children=yes`, I get the following: ``` ==3603144==...

I debugged this some more today. It turns out `_handler_wrap` isn't the problem (you can see this by replacing the handler by `ffi.NULL`, which causes R to use its own...

To see if this is a problem in other "`R_tryCatchError` -> Python callback -> back to R" scenarios in which the "back to R" step throws an error, you can...

On the Python side, the relevant part is https://github.com/python/cpython/blob/3.12/Python/ceval.c#L679-L705. Notice that `cframe` is stack-allocated by `_PyEval_EvalFrameDefault`, but pointers to it are stored on the heap inside the thread state object...