-
Notifications
You must be signed in to change notification settings - Fork 426
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
Overlinkage warnings raised for python
and numpy
#3325
Comments
Thanks @jakirkham.
This isn't a warning, it's just informative but the text is incorrect. I propose changing it to:
I personally prefer to see it, but am happy to go with the majority here (or some suggestion for better logic/user control around the info/warning/error messages in this code in general).
This one is trickier. The logic is currently:
Now this is generally speaking what we want (I contend there aren't many dependent packages with DSOs where you don't autoload any of them or load them dynamically as plugins). For Python it's only complaining because we've been kind enough to provide python DSOs for people who wish to embed our Python (even though we never use them ourselves - when this happens in an extension module it is always a bug). I do have logic to determine what constitutes a plugin package, and that basically hardcodes to them being dependent upon r-base, mro-base or python so I guess I could hardcode this logic to exclude those from the |
Hi there, thank you for your contribution! This issue has been automatically marked as stale because it has not had recent activity. It will be closed automatically if no further activity occurs. If you would like this issue to remain open please:
NOTE: If this issue was closed prematurely, please leave a comment. Thanks! |
Contrary to the title of this issue, the warnings here are about overdepending, which is a separate concern from overlinking (though in fairness, the names aren't exactly self-explanatory, aside from being underdocumented). |
Talking about numpy specifically, the following comment from @xhochy is relevant:
Rephrasing according to my (limited) understanding, that means the linkage does not happen at build time (which is when conda-through-lief would be detecting things), and therefore that warning is inevitable -- unless numpy gets special-cased, or the link check gets enhanced to detect such runtime linkage. |
Appears
conda-build
is trying to detect linkages to Python dependencies needed during the build, but it seems to be incorrectly drawing the conclusion that they are not needed as there is not a library linkage found. This happens most notably withpython
andnumpy
as shown below.Actual Behavior
ref: https://circleci.com/gh/conda-forge/python-spams-feedstock/245
Expected Behavior
No complaints about overlinkage are issued.
Steps to Reproduce
Build this recipe with one of the variants in
.ci_support
.Output of conda info
The text was updated successfully, but these errors were encountered: