scoder

There was a recent discussion on the future state in Python in https://mail.python.org/archives/list/[email protected]/thread/SZLWVYV2HPLU6AH7DOUD7DWFUGBJGQAY/ (Also https://discuss.python.org/t/type-annotations-pep-649-and-pep-563/11363)

>2. Currently, the approach to multi-mode cython code (like >`limited_api`) has been to add C-level macros to switch on and off >different behaviour. Then the c code is generated once,...

Also see #575, which describes the instance-level part of setting "default values" for attributes. Note that the values always have to be defined in the declaration, i.e. in the `.pxd`...

> In CythonFunction.c > > ``` > cf->m_self = (PyObject *) op; > ``` > > should probably be `cf->m_self = NULL` (or just not exist at all) When I...

>Is that a Cython 3.0 change? Yes. > I've only been running with tip of the 2.9 >branch to get the fixes for 3.11, which didn't seem to be in...

I think the changes for `cythonize` are fine, but the problem with `cython` is just that the error is not being printed completely at the end of `Main.py`.

> > the problem with cython is just that the error is not being printed completely at the end of Main.py. > > Should I create separate PR for this?...

> should I fix it in this PR or the fix will go elsewhere? Whatever is necessary to implement this PR can go into this PR. It's helpful to see...

I think the specific IO exceptions line `FileNotFoundError` were only added in CPython 3.4 or so. We can't use them in 2.7. (but you can keep a comment to change...

>+ccache 2>/dev/null -v -v You should probably pass the arguments before the stderr redirect. >So is there any need to do `-s` before running the tests? Is there a disadvantage?