Pablo Galindo Salgado
Pablo Galindo Salgado
> @pablogsal A gentle reminder - am I OK to merge and backport this for 3.12/3.11.1 etc? Apologies for the delay! Yeah, doing the backport is ok. I am still...
With the output so far, there is no way to know this is something in CPython or in some of the C extensions (or Cython) you are using. Could you...
> Even with a bt like this? [#98110 (comment)](https://github.com/python/cpython/issues/98110#issuecomment-1275378736) Yes, that backtrace unfortunately doesn't tell us much. The problem may have happened because some C extension is misusing the C-API...
> If I could provider a reproducer without 3rd-party libraries I would. This was found automatically as part of CI regression testing. I understand, but please understand that for us...
That is checking that the trashcan mechanism is not called recursively inside itself. This is, no object destruction triggers the trashcan while the trashcan is running.
Notice that that check is per thread state. My guess is that greenlet is doing something weird switching states in the middle of a trashcan run into sone arbitrary code...
> `destroy_chain` is static. Why is `tstate` simply not passed? Is there any reason for it? Not that I can think of, probably the evolution of the API across different...
> you see slot_tp_finalize or similar in the C stack, which I don't here. Yeah but this this a more complex case. What's happening here is that the trashcan allows...
All cherry picks will be done hours before the release so we are safe. Just tell me what you want me to add and I will make sure is there...
The following functions has been added in 3.11 and must be (maybe) documented as well: * [ ] module: zipfile added: ['Path.suffix', 'Path.stem', 'Path.suffixes', 'ZipFile.mkdir'] * [ ] module: tempfile...