conda/conda

Cross-platform, language-agnostic binary package manager

Jaime Rodríguez-Guerra

@jaimergp

I am afraid I am not aware of any such tool :/

if anything try setting CONDA_VERBOSITY=3 and explore the logs, but that's going to be time consuming

Philippe Blain

@phil-blain

Hi everyone. Not sure if the right channel but I thought I’d ask. I noticed that the Intel Distribution for Python includes conda, and is configured to use both the intel and defaults channels. Does intel have a special agreement with Anaconda or is the use of the defaults channel still subject to Anaconda repository Terms of Service ? (i.e no commercial of gvt use)

9 replies

Pat Gunn

@pgunn

Is there a good way of getting some attention to conda/conda-docs#742 ?

psortos

@psortos

Can I get some assistance on an issue I'm having with a pytorch recipe? I'm these errors ImportError: libGL.so.1: cannot open shared object file: No such file or directory and AttributeError: module 'torch.jit' has no attribute 'unused' this is the current repo conda-forge/staged-recipes#16916

Nora Loose

@NoraLoose

Hi everyone! I merged a conda-forge/gcm_filters-feedstock#2 to update the meta.yaml to show a new license and description for our package. However, on conda-forge, the license and description of our package are still in their outdated version, even though this meta.yaml file on the feedstock contains the proper information. Does anyone know what is going on? Thanks in advance for your help!

2 replies

Pat Gunn

@pgunn

I'm having trouble figuring out what's going on in some package version parsing; I'm getting an env that's not building because of glibc version issues, but as far as I can tell this isn't actually a mismatch; wondering what I'm missing.

The following specifications were found to be incompatible with your system:

  - feature:/linux-64::__glibc==2.17=0
  - feature:|@/linux-64::__glibc==2.17=0
  - coverage -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - cython -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - h5py -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - mypy -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - numpy -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - opencv -> libgcc-ng[version='>=7.3.0'] -> __glibc[version='>=2.17']
  - psutil -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - python=3.10 -> libgcc-ng[version='>=7.5.0'] -> __glibc[version='>=2.17']
  - scikit-image -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - scikit-learn -> libgcc-ng[version='>=9.3.0'] -> __glibc[version='>=2.17']
  - scipy -> libgfortran-ng -> __glibc[version='>=2.17']
  - tensorflow -> __glibc[version='>=2.17']
  - tensorflow -> cudatoolkit[version='>=11.2,<12'] -> __glibc[version='>=2.17,<3.0.a0']
  - tifffile -> libgcc-ng[version='>=7.3.0'] -> __glibc[version='>=2.17']
  - tk -> libgcc-ng[version='>=7.5.0'] -> __glibc[version='>=2.17']

Your installed version is: 2.17


scclin009(1):~/src/caiman_dev-clean$ rpm -qa glibc
glibc-2.17-324.el7_9.i686
glibc-2.17-324.el7_9.x86_64

15 replies

Tim 'mithro' Ansell

@mithro

People here might find https://twitter.com/proppy/status/1466020384063430657?s=20 interesting

Pat Gunn

@pgunn

Do entries like this describe what packages are available in the defaults channel?

https://anaconda.org/anaconda/tensorflow/files

aceenprsu

@aceenprsu

I have conda installed by rpm on Centos 7.
I created a conda environment and installed conda on the environment.
The conda provided by the rpm masks the conda in the environment.
How should I configure the conda environment so that the conda inside the environment is used instead?

Pat Gunn

@pgunn

Is there a reason you didn't update the base environment instead?

@aceenprsu Anyhow, you should be able to edit the rpm-based conda init stuff out of your environment and put a replacement with different paths in your dotfiles

All the rpm does, presumably, is edit the /etc/profile.d/ and similar standard dotfiles; have your account dotfiles undo those changes and init with your conda env.

(the term "environment" might be a bit confusing here as it could either mean a conda virtualenv-based environment, or an install of the conda codebase with its individual environment store and so on)

Pat Gunn

@pgunn

I'm running into an issue where I cannot create a new environment with conda create ... with an existing conda install (with a lot of other environments), but I can make the same environment with a completely fresh install of conda. Any idea what might cause this?

(I'm actually trying to help out a user with this problem; there may be some turnaround in poking around)

Jaime Rodríguez-Guerra

@jaimergp

what's the error? it might be due to conda versions, space limitations, permissions... plenty of possible causes

Pat Gunn

@pgunn

Version conflicts between packages.

Jaime Rodríguez-Guerra

@jaimergp

have you added extra configuration keys to your .condarc? e.g. create_default_packages, pinned_packages, aggresive_updates?

or maybe the default channels are different?

Pat Gunn

@pgunn

I asked this prematurely; I need to get access to this person's laptop to see what's going on. Probably really inefficient to try to manage those timescales and gitter. Thanks for the ideas though.

Jaime Rodríguez-Guerra

@jaimergp

No problem! Ping me if needed!

Faustin Carter

@FaustinCarter

When running conda-build on Windows, I'm getting a bunch of new output related to LIEF. Is there a way to skip these ( other than downgrading to an earlier version of conda-buidl?)

Christian Roth

@croth1

Can anybody help me understand how to read this solving config error?

Package cudnn conflicts for:
pytorch-lightning=1.4.5 -> pytorch[version='>=1.6'] -> cudnn[version='>=7.6,<8.0a0|>=7.6.5.32,<8.0a0|>=8.2.1.32,<9.0a0|>=8.1.0.77,<9.0a0']
pytorch::torchvision=0.10.0 -> pytorch==1.9.0 -> cudnn[version='>=7.6.5.32,<8.0a0|>=8.2.1.32,<9.0a0']The following specifications were found to be incompatible with your system:

  - feature:/linux-64::__glibc==2.27=0
  - nvidia::cudatoolkit=11.1.1 -> __glibc[version='>=2.17,<3.0.a0']
  - pytorch-lightning=1.4.5 -> pytorch[version='>=1.6'] -> __glibc[version='>=2.17|>=2.17,<3.0.a0']
  - pytorch::pytorch=1.9.0 -> cudatoolkit[version='>=11.1,<11.2'] -> __glibc[version='>=2.17,<3.0.a0']
  - pytorch::torchvision=0.10.0 -> cudatoolkit[version='>=11.1,<11.2'] -> __glibc[version='>=2.17|>=2.17,<3.0.a0']

Your installed version is: 2.27

Why would __glibc==2.27 conflict with __glibc[version='>=2.17|>=2.17,<3.0.a0']? What am I overlooking here?

Jaime Rodríguez-Guerra

@jaimergp

Might be a red herring. If pytorch is involved (which requires 2.17 as you see), there's a chance a package for 2.12 is involved somehow in the dependency tree. Try solving with mamba and see if their error is a bit clearer.

1 reply

Marcelo Duarte Trevisani

@marcelotrevisani

Could someone please review this PR?
conda/conda#11123

Maximilian Koch

@tschaka1904

Hi all,

I'm trying to build a package which includes h5py. When using conda build, it seems to install the wrong version of the dependency. It installs 3.2.1-py37h6c542dc_0, which includes hdf5: 1.10.6-nompi_h6a2412b_1114. The problem is that this hdf5 lib, seems to have these setting:
(Read-Only) S3 VFD: yes

This causes an error for me. When just running conda install h5py==3.2.1, it does install the right version.

Chris Meyer

@cmeyer1969:matrix.org

[m]

Hi all, I noticed that packages for Python 3.10 are starting to become available. However, some packages (like h5py) are not yet available on Windows. h5py depends on pkgconfig, which is also not available on Windows. Are these packages expected to be available currently or are they still being built or is there some other reason they are not available on Windows?

Pat Gunn

@pgunn

Could whomever maintains the anaconda/miniconda docker images consider adding mamba to those images? It's pretty important for CI use.

7 replies

Isuru Fernando

@isuruf

@jezdez, can we speed up the conda-build issue/PR locking? I get around 5 emails everyday for about 5 days now and it's annoying

2 replies

cc @chenghlee

Fred Eisele

@fredrick.eisele_gitlab

Is there a way to promote a conda environment to configure the default os environment?
I am not referring to the conda base environment.

Benjamin Bertrand

@beenje

conda environments are isolated by nature. What I have done in the past is to automatically activate a specific environment using a file under /etc/profile.d on Linux so that users can use that env directly. For command line tools, another solution I used is to create simple wrappers under /usr/local/bin. Users can launch a tool without knowing it's installed in a conda env.

Ian Stokes-Rees

@ijstokes:matrix.org

[m]

I'm getting repeated HTTPS connection errors from conda when trying to do conda update conda (currently running 4.11.0) -- using --debug the problem seems to be [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate -- is this just me or is there some kind of server-side issue? If it is just me can someone suggest what I need to do to correct this?

1 reply

Pat Gunn

@pgunn

I'm wondering if there's an existing tool to relocate a conda directory, before I get too deep into writing one for our particular needs

Right now it looks like the user's conda install path gets written into a lot of things under each environment's BASE/bin, the main CONDA_BASE/bin, as well as ~/.conda/environments.txt

Pat Gunn

@pgunn

It looks like conda-pack does a subset of what I want, but I was hoping for something that rewrites paths for every environment

Jannis Leidel

@jezdez

I think https://github.com/conda/conda-prefix-replacement may do that?

Andrew Kapocsi

@akapocsi

The link to the conda mailing list in the README of the conda repo is dead: https://github.com/conda/conda#getting-help

Anyone know if there is a replacement?

Leo Fang

@leofang

Any chance CDN sync has not happened in the past ~4 hrs? I am seeing packages that were uploaded a few hrs ago are not yet visible to the conda solver.

1 reply

daniel.bast

@daniel.bast:matrix.org

[m]

the new multi-platform conda-ci container enable much faster test executions on Apple M1:

time docker run --platform linux/amd64 --rm -v ${PWD}:/opt/conda-src -e TEST_SPLITS=3 -e TEST_GROUP=1 ghcr.io/conda/conda-ci:master-linux-python3.9 /opt/conda-src/dev/linux/unit.sh needs 9:17.24
time docker run --platform linux/arm64 --rm -v ${PWD}:/opt/conda-src -e TEST_SPLITS=3 -e TEST_GROUP=1 ghcr.io/conda/conda-ci:master-linux-python3.9 /opt/conda-src/dev/linux/unit.sh needs 2:41.42 -> factor 3.4 faster :)

1 reply

Mikey

@BrainSlugs83

Getting 403 errors trying to install pytorch packages with conda -- is there any known outages right now? --
The packages are listed in the directory listing for the server, but clicking on them gives an XML error and 403 message saying "Access Denied" and "Request has Expired":

One of the failing packages: https://conda.anaconda.org/pytorch/noarch/pytorch-mutex-1.0-cuda.tar.bz2
Linked Bug on github: pytorch/pytorch#81798

Mark Harfouche

@hmaarrfk

There was a known outage. Not sure if it is resolved. Definitely fewer error this morning.

https://www.cloudflarestatus.com/

might be related.

Cheng H. Lee

@chenghlee

It was cloudflare-related but nothing to do with the outages posted on their status page. In any case, Anaconda pushed a hotfix to anaconda.org that should mitigate the issue.

1 reply

mattalhonte-srm

@mattalhonte-srm

Heya! keeps trying to install different Python versions and then failing when it doesn't find some dep - example error: - nothing provides openssl >=1.1.1,<1.1.2.0a0 needed by python-3.7.1-h0371630_3
I explicitly put python==3.8.* (also tried it with a single = in my yaml file.
Not really sure what to do? Is there a reliable way to pin the Python version? Is there a reliable way to do detective work and see what package in my yaml file is responsible for this chain of events?
Thanks!

Stephen Nayfach

@snayfach

I updated my package checkv yesterday (https://anaconda.org/bioconda/checkv) but the new version (1.0.0) still is not available when I run conda install -c bioconda checkv. Is there something I did wrong when I updated the recipe?

1 reply

Where communities thrive

People

Repo info

Activity