0

Today I tried to upgrade my conda installation and came across a strange behaviour.

The upgrade to conda 22.11. from 22.9. did not work. After updating (environment solving works, files are downloaded and upgraded) changes are made but conda stays at 22.9.0

A repeated update attempt yields:

conda update -n base -c defaults conda

Collecting package metadata (current_repodata.json): done

Solving environment: done

==> WARNING: A newer version of conda exists. <== current version: 22.9.0 latest version: 22.11.1

Please update conda by running

$ conda update -n base -c defaults conda

All requested packages already installed.

Any help? Any ideas? Thank you very much

I expect a clean update of conda to 22.11.

According to the specs. conda 22.11. should still run with Python 3.7 (?)

merv
  • 67,214
  • 13
  • 180
  • 245
Robert S
  • 1
  • 1

2 Answers2

0

Yes, anaconda (or defaults) channel still builds for Python 3.7 (note that conda-forge does not). In my opinion, the update subcommand is underpowered and lacks expressivity. If you know you want a particular version, then the install subcommand fully supports MatchSpec and has more imperative force. That is, try something like:

conda install -n base -c defaults 'conda>=22.11'

which will either get Conda to install it or at least trigger it to try to explain why it cannot.

merv
  • 67,214
  • 13
  • 180
  • 245
  • Thank you merv, for you suggestion. I tried, but conda would run into a problem solving the environment. Upon trying with a "frozen" environment it took over/night to solve and then I got pages and pages of incompatibilities. – Robert S Jan 17 '23 at 10:00
0

Thank you Merv for your suggestion. I tried a forced update of conda and it would fail to solve the environment. Upon solving in a "frozen" environment (which took the whole night) it would come up with pages and pages of incompatibilities.

When checking what revisions had been done originally during the general update I came across a likely explanation:

`2023-01-13 10:36:34  (rev 11)
     _libgcc_mutex  {0.1 (defaults/linux-64) -> 0.1 (conda-forge/linux-64)}
     asn1crypto  {1.3.0 (defaults/linux-64) -> 1.3.0 (conda-forge/linux-64)}
     attrs  {19.3.0 (defaults/noarch) -> 19.3.0 (conda-forge/noarch)}
     backports  {1.0 (defaults/noarch) -> 1.0 (conda-forge/noarch)}
     backports.functools_lru_cache  {1.6.4 (defaults/noarch) -> 1.6.4 (conda-forge/noarch)}
     backports.tempfile  {1.0 (defaults/noarch) -> 1.0 (conda-forge/noarch)}
     backports.weakref  {1.0.post1 (defaults/noarch) -> 1.0.post1 (conda-forge/noarch)}
     bokeh  {1.4.0 (defaults/linux-64) -> 1.4.0 (conda-forge/linux-64)}
     chardet  {3.0.4 (defaults/linux-64) -> 3.0.4 (conda-forge/linux-64)}
     docutils  {0.16 (defaults/linux-64) -> 0.16 (conda-forge/linux-64)}
     flake8  {3.7.9 (defaults/linux-64) -> 3.7.9 (conda-forge/linux-64)}
     future  {0.18.2 (defaults/linux-64) -> 0.18.2 (conda-forge/linux-64)}
     glob2  {0.7 (defaults/noarch) -> 0.7 (conda-forge/noarch)}
     heapdict  {1.0.1 (defaults/noarch) -> 1.0.1 (conda-forge/noarch)}
     imageio  {2.6.1 (defaults/linux-64) -> 2.6.1 (conda-forge/linux-64)}
     importlib_metadata  {1.5.0 (defaults/linux-64) -> 1.5.0 (conda-forge/linux-64)}
     intervaltree  {3.0.2 (defaults/noarch) -> 3.0.2 (conda-forge/noarch)}
     jdcal  {1.4.1 (defaults/noarch) -> 1.4.1 (conda-forge/noarch)}
     jsonschema  {3.2.0 (defaults/linux-64) -> 3.2.0 (conda-forge/linux-64)}
     jupyter_client  {5.3.4 (defaults/linux-64) -> 5.3.4 (conda-forge/linux-64)}
     nbconvert  {5.6.1 (defaults/linux-64) -> 5.6.1 (conda-forge/linux-64)}
     nose  {1.3.7 (defaults/linux-64) -> 1.3.7 (conda-forge/linux-64)}
     pathtools  {0.1.2 (defaults/noarch) -> 0.1.2 (conda-forge/noarch)}
     pexpect  {4.8.0 (defaults/linux-64) -> 4.8.0 (conda-forge/linux-64)}
     pip  {20.0.2 (defaults/linux-64) -> 20.0.2 (conda-forge/linux-64)}
     ptyprocess  {0.6.0 (defaults/linux-64) -> 0.6.0 (conda-forge/linux-64)}
     pyopenssl  {19.1.0 (defaults/linux-64) -> 19.1.0 (conda-forge/linux-64)}
     pysocks  {1.7.1 (defaults/linux-64) -> 1.7.1 (conda-forge/linux-64)}
     pytest-astropy-header  {0.1.2 (defaults/noarch) -> 0.1.2 (conda-forge/noarch)}
     python-dateutil  {2.8.1 (defaults/noarch) -> 2.8.1 (conda-forge/noarch)}
     pytz  {2019.3 (defaults/noarch) -> 2019.3 (conda-forge/noarch)}
     requests  {2.22.0 (defaults/linux-64) -> 2.22.0 (conda-forge/linux-64)}
     six  {1.14.0 (defaults/linux-64) -> 1.14.0 (conda-forge/linux-64)}
     sphinxcontrib-jsmath  {1.0.1 (defaults/noarch) -> 1.0.1 (conda-forge/noarch)}
     sympy  {1.5.1 (defaults/linux-64) -> 1.5.1 (conda-forge/linux-64)}
     traitlets  {4.3.3 (defaults/linux-64) -> 4.3.3 (conda-forge/linux-64)}
     wheel  {0.34.2 (defaults/linux-64) -> 0.34.2 (conda-forge/linux-64)}
     xmltodict  {0.12.0 (defaults/noarch) -> 0.13.0 (conda-forge/noarch)}

`

A lot of dependencies had moved to conda-forge, and that, according to your info does not support Python 3.7 any more. That appears to be the likely reason.

As a remedy I reverted to the old revision 10. It is unfortunate that the update routine does not give one a proper warning. I guess no more conda updates on my machine (have a lot of Python 3.7 scripts running).

Thank you again.

Robert S
  • 1
  • 1