Pythonise the docs #4187: "Extension types (aka. cdef classes)" (cdef_classes.rst)

[0dminnimda]

Continuing to Pythonize Сython documentation!

This PR Pythonizes this documentation page.

Lots of process details and standardization are found here: #4187

BTW, the fact that this is a draft PR does not mean that you can no longer leave review, as I understand it, so I urge you to do it.

[scoder]

I think a note is best. Once readers have seen it on one page, it's easy to ignore on others. Highlighting the paragraph in a Note makes it clear that it's not part of the main page content.

[0dminnimda]

@scoder @da-woods @Bluenix2 @mattip Faced, as I understand it, a kind of this (#1428) issue. Described the behavior here. (Not sure if there is enough difference to create a new issue)

The problem is that if people follow this tutorial step by step, they will face this behavior.
In general, the development process with pure python and using pxd is very inconvenient due to this problem.

If we want to show people pure Python code using pxd, they should definitely be warned about this issue.

It seems that this rather unpleasant problem can be solved quite simply, as far as I understand, there is already progress on its solution: #4063

Sorry for sending notifications, I just think if you can, it would be good to try to think about a solution / review the proposed PR.

[da-woods]

If we want to show people pure Python code using pxd, they should definitely be warned about this issue.

"Creating the .pxd file may not cause Cython to re-compile the .py file. If you are using setup.py to build you can force recompilation with the --force argument."?

It seems that this rather unpleasant problem can be solved quite simply, as far as I understand, there is already progress on its solution: #4063

I think the solution is pretty close to complete - it just needs the testing tidied up a bit. It's probably worth trying to get it merged fairly soon.

[0dminnimda]

it just needs the testing tidied up a bit.

It seems that the author of that PR is not in the business of improving tests.

Oh yes, you, as a cython collaborator, can make commits in PR.

Is there anything I can do to help improve those tests?

[0dminnimda]

If you are using setup.py to build you can force recompilation with the --force

Yes, it looks like a temporary solution, but it is still unpleasant to compile, for example, a large number of files every time you change one.

[0dminnimda]

@scoder do you have any comments?
The only thing that might be helpful is to add a warning about #1428
Anyway, the next thing I will do after merging this PR is to solve the above issue.

[0dminnimda]

@scoder #1428 is now closed so there is no need to add a warning anymore. Can this PR be merged now?

[0dminnimda]

@scoder Maybe we will merge this PR?

[0dminnimda]

oops experimented a little with fork branches

[0dminnimda]

@scoder please merge this PR.

[scoder]

Thanks

[0dminnimda]

Thanks

Thanks you, finally this is merged ..