-
Notifications
You must be signed in to change notification settings - Fork 3k
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
pip 20.2b1 resolver still failing with ResolutionTooDeep for jupyter install #9997
Comments
There have been many improvements since 20.2b1. Is this reproducible with the latest pip version (21.1.1)? |
Correct hit the same issue in 21.1.1 and worked back to 20.2.b1 to see if this was a solution, but hit the same limit. In the end solved the problem by backing off docker image's jupytherhub to 1.0.0. |
Do you still have this issue in pip 21.3.1? A significant backtracking optimization landed in pip 21.3, and the ecosystem has largely improved. I couldn't reproduce your example exactly as it seems to depend on internal packages I don't have access to. |
@notatallshaw The packages "tpsubmitter" and "tpscripts" are internal, and I do not believe necessary to reproduce this error. Can you try striking them, and changing the docker image to "jupyterhub:1.4.0"? If that does not produce the error, let me know. I am not doing python development now so hoping you can reproduce with this modification. If not let me know and I will have to recreate my python development env and work the issue along with you. |
Your OP already uses "jupyterhub:1.4.0" right? Following the instructions as is I get an error related to building pandas. I was able to fix this by adding these lines before installing the requirements file:
After that the docker file built fine. |
This should now be resolved on Pip main (28239f9) due to sarugaku/resolvelib#111 and sarugaku/resolvelib#113. I haven't been able to reproduce any other long backtracking issue with these improvements. Can you retry by installing Pip like so: |
You can now install Pip 23.1 which has massively improved backtracking capability, I was never able to reproduce your issue but if you have clear instructions that didn't involved internal packages I would be happy to try. |
Given that we haven't reproduced this in over an year and with the recent improvements, I'll close this out. |
Description
Building jupyter exceeds the incredibly large number of 2,000,000 (Set max_rounds to an incredibly large number #8275) for the pip 20.2b1 dependency resolver recommended in (pip resolver fails in a fresh env when installing jupyter with other stuff #8290).
Docker Image: docker:18.09.7 / docker:18.09.7-dind
Pip: pip 20.2b1 from /usr/local/lib/python3.8/dist-packages/pip (python 3.8)
Python: Python 3.8.5
Expected behavior
It would nice if we didn't have 2 million dependencies evaluated. I am all for an improved dependency resolver, but this is not an improvement for this case. I had to roll back to juypterhub 1.0.0 to circumvent this.
pip version
20.2b1
Python version
3.8.5
OS
Docker Image: docker:18.09.7 / docker:18.09.7-dind
How to Reproduce
Dockerfile
requirements.txt
Output
Code of Conduct
The text was updated successfully, but these errors were encountered: