Robert Bradshaw

We don't have a specific roadmap, but given that we're a low-level dependency we do have a history of supporting Python versions long after EOL. We are working on changing...

+1 On Wed, Mar 27, 2019 at 7:16 PM Stefan Behnel wrote: > Given that Py3.4 is now out of maintenance, and that the test suite takes > excessively long...

I would be in favor of deprecating as well. Some `IF` expressions that would not be legal if replaced by if are those that use, for example, nonexistent or differently...

I'd welcome extensions to produce an HPy backend. I too would prefer generating a single `.c` file with both modes, continuing to use macros to target different backends. I think...

Correct, this is not yet supported. For now, the best way to do this is simply to use a module-level global (and possibly do an assignment to a default value...

I think it's clear what the intent is: tuples that are part of a unit for unpacking purposes (and arguably lambda arguments fall in the same boat) should bind more...

Thanks for the report. Indeed, the code is correct, but the C compiler does not understand the subtype relationship among these pointers (hence the warning and error).

Sweet, this works. I'll have to admit I'm a bit confused tough--is the sender always sending to the default port no matter the port parameter passed?

So would users be expected to import these transforms from the `_et` paths directly? One transform per module (which is not how the others are typically arranged, e.g. with the...

Ah, yes, that's right. Hopefully the file generation itself isn't *that* slow. On Mon, Feb 12, 2024 at 3:02 PM Ahmed Abualsaud ***@***.***> wrote: > ***@***.**** commented on this pull...