0

I work on a machine that has anaconda 4.10.3 pre-installed. I am unable to update it to a newer version. The problem with this version is that it interprets "Python 3.10.2" as "Python 3.1". It cuts of the version number after the first character behind the dot. When I try to install pytorch I get the following error:

PackagesNotFoundError: The following packages are not available from current channels: 

- python=3.1

Is there a possibility to ignore this error and just continue with the installation?

Ulrich Eckhardt
  • 16,572
  • 3
  • 28
  • 55
  • 1
    Seems like a bug, but not in any code you wrote or show here, so I'd say it's off-topic. Please file a bug ticket. Still, read [mcve], because a good bug ticket contains that as well. – Ulrich Eckhardt Feb 28 '22 at 19:26
  • @UlrichEckhardt All I ask is if there is a possibility to ignore this conda error and continue with the pytorch installation. Something like `--force`. – abcdefghi999955 Feb 28 '22 at 19:37
  • @UlrichEckhardt it is a known bug whose fix is to upgrade to Conda 4.11+, which OP says is not possible. – merv Feb 28 '22 at 22:47
  • What is your command to install PyTorch? Essentially, you just can never use a Python 3.10+ with this. Have you tried: `conda create -n foo -c pytorch python=3.9 pytorch`? Otherwise, I really don't think you can *ignore*; simply have to always specify a Python version. – merv Feb 28 '22 at 22:49
  • Create a wrapper around Python that reports version 3.9 when asked? Downgrade Python? Backport the fix to the old version? Disable the check check entirely? Get your admin to do these things for you? Sorry, but this is not a programming issue (that problem was solved long ago!) but one with your environment and what you can or can't do there. – Ulrich Eckhardt Mar 01 '22 at 08:27
  • @merv: If it's a known bug, where is the link to the bug? Could you add that to the question? Are there any workarounds in the bug ticket perhaps? Are there possibly already multiple questions about Python >=3.10 and Conda <4.11 here already? – Ulrich Eckhardt Mar 01 '22 at 08:29
  • Even though this is a slightly different question, I’m closing because workarounds are already documented in previous question. And, no, the error cannot be ignored. As Ulrich suggests, file a request with your IT - they should upgrade Conda. – merv Mar 01 '22 at 13:48

0 Answers0