-
Notifications
You must be signed in to change notification settings - Fork 14
Update Python to 3.13.2 and fix compilation of x86_64 cpython extensions #253
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
The "build" package is becoming the main Python package build frontend. As such, it should be in basement because it is part of the standard Python build support.
This is the main class to use Python packages at build time. It is used by all other Python infrastructure. Thus give it a common, short name.
Put all classes that handle the different types of Python build systems into a common namespace. At the same time, split the setuptools and pip methods from one another. The python3::cext module is converted to a mixin class. Whether or not a Python module has a C-extension is independent from the underlying build system.
So far, the integrated distutils package will add the Python LIBDIR unconditionally to the linker command line. This is bad because this path points to some global path e.g., /usr/lib! It is even unnecessary because the linker should find the python shared object anyway already.
0c1b7c7 to
83a4ed3
Compare
|
@rhubert Could you have a look if this breaks anything on your side? You'll also have to use BobBuildTool/basement-gnu-linux#78. I know it creates some churn but I was uncomfortable with the classes already in the past and the pip update finally pushed it over the cliff for me... |
|
I tried it and my problems are solved with the latest cl. |
|
I tried to build with these changes but it fails with (dev-sandbox): I don't see how this relates to each other but it worked before... 🤔 |
|
This looks indeed unrelated. I made a clean build and it did not show the error. My strong suspicious is that the relevant patch ( I checked again the patch class but could not see how we can end up with such a situation. Any chance you still know what steps lead to the error? |
Thanks, that's indeed the case. But I've no Idea what happened. The log is: The last two builds are very close to each other. I can't remember what I did two days ago. 1s seams to be even to short for interrupting + restart. Maybe some bob-cmd.log would help. 🤔 Anyway - now I got further up to ninja - which seams to need a update now: |
|
The archive was indeed extracted again. Bummer. 🤯 I've added b0a61100e2ca392956498652212744d1e3155314. Even though it is highly unlikely, the floating point comparison is not nice. Let's hope it was a one-off freak event... 😉 |
@sixtyfourktec This should fix compilation of x86_64 python extension modules...