Antonio Cuni

> > Nothing prevents python implementation to provide their own specific headers, of course and/or to decide that it will NOT support universal modules. > > I think you mean...

> My point was more that with HPyTuple_Check, we have a place (a proper API call) where the implementation can attach that type information to the object behind the returned...

> What if someone creates a native subclass of tuple? These will have to be allocated using malloc, so that's just some native allocation. Will we force them to do...

> I am mostly concerned with macros that access struct members, which I would like to avoid, because in our case that often means a wrapper or exposing that member...

I must say that I'm not actively involved with any pdbpp development at all these days: I think they only reason why I'm still listed as maintainer is for historical...

I agree that you should show hidden folders by default. But these are special, because they are not present in the static website: if you click on them you see...

Some more thoughts about this. My initial idea is that once we have greenlets, it should be easy/possible to simulate synchronous calls from Python. E.g., I can imagine having an...

Uhm, I admit that I don't fully understand *everything* here and I am probably missing some pieces. But I wonder whether the approach used by PyPy can be useful here....

So if I understand correctly, pyodide's greenlets will be implemented on top of continulet as we do on PyPy?

> This is the plan. @antocuni Can you provide any extra information about what the `permute` function is supposed to do? I don't really understand how it is supposed to...