Skip to content
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

Packages that are only available as .conda artifacts are not reliably found #2172

Closed
michaelosthege opened this issue Dec 16, 2022 · 10 comments

Comments

@michaelosthege
Copy link

I tried mamba create -n testenv -c conda-forge "sunode==0.4.0" and got nothing provides numpy >=1.23.5,<2.0a0 needed by sunode-0.4.0-py310hd5a9e38_1.

This prompted me to try mamba create -n testenv -c conda-forge "numpy==1.23.5" resulting in nothing provides requested numpy 1.23.5.

My OS is Windows x64 and I did this with latest mamba v0.27.0.

Earlier today I had similar problems with pymc==5.0.0 and sunode==0.4.0 and got them resolved by updating the feedstocks to release not only .conda artifacts (the new default), but also .tar.bz2 artifacts as was the default until a month or so ago.

conda_build:
  pkg_format: 2    # makes .conda artifacts
  pkg_format: None # makes .tar.bz2 artifacts

Strangely enough, this fixed the install of pymc==5.0.0, sunode==0.4.0, numpy==1.23.5 for my colleague (also on Windows).

Since NumPy 1.23.5 is currently only listed with .conda artifacts, and it worked for my colleague but not for me,
this falsifies my theory that it was a matter of having .tar.bz2 artifacts.

@beckermr
Copy link

Did you try conda?

A useful debugging data point and a possible workaround.

@michaelosthege
Copy link
Author

michaelosthege commented Dec 16, 2022

Did you try conda?

A useful debugging data point and a possible workaround.

Yes, did that two minutes ago in a Python 3.9 env created with mamba create -n testenv -c conda-forge "python=3.9"

>>> conda install -c conda-forge "numpy==1.23.5"
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.

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

  - numpy==1.23.5

Current channels:

  - https://conda.anaconda.org/conda-forge/win-64
  - https://conda.anaconda.org/conda-forge/noarch
  - https://repo.anaconda.com/pkgs/main/win-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/r/win-64
  - https://repo.anaconda.com/pkgs/r/noarch
  - https://repo.anaconda.com/pkgs/msys2/win-64
  - https://repo.anaconda.com/pkgs/msys2/noarch

with

>>> conda --version
conda 4.12.0

@beckermr
Copy link

Works fine for me

$ conda create -n testenv -c conda-forge "numpy==1.23.5"

***

NOTE: You are using the EXPERIMENTAL libmamba solver integration.

If something is not working as expected, please:

1. Go to https://github.com/conda/conda/issues/new/choose
2. Choose the "Libmamba Solver Feedback (Experimental Feature)" option

Thank you for your help!

***

pkgs/r/osx-64                                                 No change
pkgs/main/noarch                                              No change
pkgs/r/noarch                                                 No change
pkgs/main/osx-64                                     4.8MB @   3.5MB/s  1.6s
conda-forge/noarch                                  10.6MB @   3.5MB/s  3.5s
conda-forge/osx-64                                  25.9MB @   4.3MB/s  7.4s
Collect all metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: /Users/beckermr/mambaforge/envs/testenv

  added / updated specs:
    - numpy==1.23.5


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    libgfortran-5.0.0          |11_3_0_h97931a8_27         143 KB  conda-forge
    libgfortran5-11.3.0        |      h082f757_27         1.4 MB  conda-forge
    numpy-1.23.5               |  py311h62c7003_0         6.3 MB  conda-forge
    python_abi-3.11            |          3_cp311           6 KB  conda-forge
    ------------------------------------------------------------
                                           Total:         7.8 MB

The following NEW packages will be INSTALLED:

  bzip2              conda-forge/osx-64::bzip2-1.0.8-h0d85af4_4 None
  ca-certificates    conda-forge/osx-64::ca-certificates-2022.12.7-h033912b_0 None
  libblas            conda-forge/osx-64::libblas-3.9.0-16_osx64_openblas None
  libcblas           conda-forge/osx-64::libcblas-3.9.0-16_osx64_openblas None
  libcxx             conda-forge/osx-64::libcxx-14.0.6-hccf4f1f_0 None
  libffi             conda-forge/osx-64::libffi-3.4.2-h0d85af4_5 None
  libgfortran        conda-forge/osx-64::libgfortran-5.0.0-11_3_0_h97931a8_27 None
  libgfortran5       conda-forge/osx-64::libgfortran5-11.3.0-h082f757_27 None
  liblapack          conda-forge/osx-64::liblapack-3.9.0-16_osx64_openblas None
  libopenblas        conda-forge/osx-64::libopenblas-0.3.21-openmp_h429af6e_3 None
  libsqlite          conda-forge/osx-64::libsqlite-3.40.0-ha978bb4_0 None
  libzlib            conda-forge/osx-64::libzlib-1.2.13-hfd90126_4 None
  llvm-openmp        conda-forge/osx-64::llvm-openmp-15.0.6-h61d9ccf_0 None
  ncurses            conda-forge/osx-64::ncurses-6.3-h96cf925_1 None
  numpy              conda-forge/osx-64::numpy-1.23.5-py311h62c7003_0 None
  openssl            conda-forge/osx-64::openssl-3.0.7-hfd90126_1 None
  python             conda-forge/osx-64::python-3.11.0-h559f36b_0_cpython None
  python_abi         conda-forge/osx-64::python_abi-3.11-3_cp311 None
  readline           conda-forge/osx-64::readline-8.1.2-h3899abd_0 None
  tk                 conda-forge/osx-64::tk-8.6.12-h5dbffcc_0 None
  tzdata             conda-forge/noarch::tzdata-2022g-h191b570_0 None
  xz                 conda-forge/osx-64::xz-5.2.6-h775f41a_0 None


Proceed ([y]/n)? 

Maybe that new version of numpy dropped python 3.9?

@jonashaag
Copy link
Collaborator

jonashaag commented Dec 16, 2022

What are the differences between your colleague's setup and yours? Do you have the same .condarc? In particular, do you use the same channels? Are you using some corporate mirror?

@michaelosthege
Copy link
Author

michaelosthege commented Dec 16, 2022

Works fine for me

You appear to be on osx; I'm on Windows

Maybe that new version of numpy dropped python 3.9?

No, this is the link to the .conda artifact for NumPy 1.23.5 under Python 3.9 on Windows: https://anaconda.org/conda-forge/numpy/1.23.5/download/win-64/numpy-1.23.5-py39hceebc26_0.conda

My .condarc:

unsatisfiable_hints: true

He has

channels:
  - conda-forge
  - defaults
channel_priority: strict

Our mamba info differs mainly in the conda version and directory paths: https://www.diffchecker.com/YhLiY8r8

My colleague is on conda version 22.9.0 under Python 3.9.13 while I'm on conda version 4.12.0 and Python 3.7.2.

Maybe I need to update my conda and base environment Python versions?
What is a safe way to do that?

EDIT: Updating my conda with mamba update -c conda-forge conda to 22.9.0 didn't help.

@jonashaag
Copy link
Collaborator

If you could reproduce this in a Docker container that would be great

@michaelosthege
Copy link
Author

Interestingly I can create the environments on my personal computer with conda 22.9.0, python 3.8.15 and mamba 1.1.0.

I suspect the Python version of the base environment, because the 3.7 on my work PC is quite old.
I'll to to get it updated and see if that helps.

@jonashaag
Copy link
Collaborator

If you could reproduce this in a Docker container that would be great

I just realized that this wasn’t helpful because you can’t build Windows Docker containers (right?)

@jonashaag
Copy link
Collaborator

I don’t think updating the base Python will have any effect. As long as you can run a recent version of Mamba you should be able to install .conda packages.

Try removing most channels from your condarc (eg keep only conda-forge)

@michaelosthege
Copy link
Author

0.27.0 is the latest mamba version that runs with a Python 3.7 base environment.

Since this would create problems eventually, I decided to upgrade my base environment - by removing all envs, uninstalling miniconda and then installing the current miniconda.

And it helped: With a current miniconda running a Python 3.9 base, and mamba 1.1.0 I can finally create the environments 🥳

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants