scoder

We are only adding Cython specific macros to the public header files here, and including them from Cython modules will just let them do the same thing that those modules...

> I think we should guard the whole section against redefinition when included in Cython modules Thinking about this a bit more, adding a new macro guard isn't safe either...

A major goal of supporting `sys.monitoring`, from the PoV of Cython, was to get over the need for clumsily emulating frame state. Python interpreter frames are not part of the...

So, is there a test missing for this?

3.2.x commit: https://github.com/cython/cython/commit/ce82a3232afa9048b4d18d9614ebc36125330403

> the return value isn't very well identified here. I'd rather rely on keyword arguments for the readability. We can easily require some kind of strict spelling, even if Python...

Time for bikeshedding. Alternatives for `except_plus`: - `exceptval='+'` (It cannot be combined with other exception values anyway.) - `except_cpp=True` Since `except +` is not currently needed for _implementing_ functions, I...

> `cython.pointer[cython.function_type([cython.int], cython.double, except_val=-1)] # double (*)(int) except? -1` This is not completely the same. `except_val=-1` should correspond to `except -1`. I propose to add `check_exception=True` for the `except?` case....

> `exceptval='+'` might be confused for `except '+'`. But if exception values won't ever be strings, we can use this. We could probably run into user code wanting to do...

> `cython.pointer[cython.function_type([cython.int], cython.double, check_exception=False)] # double (*)(int) noexcept` This should map to `double (*)(int)`. I don't remember right now if we defaulted this to `except -1.0` or an implicit `noexcept`....