Closed mks2183 closed 1 year ago
@mks2183 I think the issue is related to some MacOS downloaded file. I don't know much about the CI, but maybe we need to create a dedicated mirror to avoid these frequent errors on MacOS.
@mks2183 I think the issue is related to some MacOS downloaded file. I don't know much about the CI, but maybe we need to create a dedicated mirror to avoid these frequent errors on MacOS.
As I have said before this error has nothing to do with downloading a file. Why do you keep saying that. There is no world where we should be mirroring pypi.
You are right @btashton, this is not related to download timeout or similar as in the past, seems like some modification in the pypi files itself. But having a controlled mirror we could make the packages fixed or only update when after testing some staging CI before updating to use in the main CI.
This is a cython 3.0 issue, it is broken within python 3.12 ; till today the only solution that is to get a prev version install, meanwhile, I agree relaying on pip is tricky, I have that experience to bind a lot of numerical / ML /algebra algorithms from C to python using cython, I banned that package manager long ago, I do maintain my own toolchain separated , why? because it randomly breaks packages (i.e working a lot with Numpy, SciKit cython interfaces... can't afford random, already enough difficult and challenging by itself without adding up an extra layer of unexpected events: pip
).
Hi Sir @acassis / @btashton,
Sir @btashton , I am totally unaware of the issue but issue with pypi as per first instance as mentioned by Sir @acassis and you told. Sorry for my end if it was discussed earlier as I am not totally aware of this.
Thanks Sir, for clarification.
Regards, Manish
Anyway, the ci is broken now. I will disable macOS ci tomorrow if no body can fix it soon.
Hi All,
I am seeing error in build where PR #9846. Is it something related with macOS ? Did anybody face it ?
Regards, Manish