Where communities thrive

  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
Repo info
  • 11:19
    vs1612 opened #10816
  • 11:11
    vs1612 commented #10794
  • 11:10
    vs1612 closed #10794
  • 09:25
    j-neefs commented #7279
  • 06:40
    caifazhou commented #8926
  • 04:59
    drorata opened #10815
  • 01:12
    Aanchal26081 commented #9746
  • Jul 27 18:36
    Ebernn commented #9746
  • Jul 27 16:42
    agilgur5 commented #10333
  • Jul 27 16:42
    agilgur5 commented #10333
  • Jul 27 14:02
    1900011604 opened #10814
  • Jul 27 09:10
    loikein commented #9707
  • Jul 27 08:08
    1900011604 edited #10813
  • Jul 27 08:07
    1900011604 opened #10813
  • Jul 27 07:04
    jonashartwig opened #10812
  • Jul 27 03:46
    adoug commented #9367
  • Jul 26 16:00
    Guillaume-Garrigos commented #9395
  • Jul 26 15:40
    hoyasam commented #10111
  • Jul 26 12:33
    janxkoci commented #10056
  • Jul 26 08:59
    jaimergp commented #10811
Wolf Vollprecht
e.g. if i have ros-pkg1 and ros-pkg2 and both depend on Python in the host section, then Python should be exactly the same in both cases
so if one first removes all packages that are part of the build from the dependencies, and then creates one big (fake) host section, then solves, then one should have the correct deps for all packages
this would cut the build time for our ROS meta.yamls by an hour easily :) so I really need to get to that
Matthew R. Becker
at some point conda build was caching envs iiuic but this led to issues
Wolf Vollprecht
@kalefranz I have compiled a short list of low-level things that I think could be improvements to conda: https://docs.google.com/document/d/1nNvHbCmvK9lNzRd28Ki3rFHSEiXzPaR7vP6830tbXlA/edit?usp=sharing
I think gradually we'll try to tackle them in mamba
maybe you guys want to chime in?
Pat Gunn
Is there a way to tell what packages are in a conda environment without running any conda commands?
Wolf Vollprecht
conda list?
or check te files in conda-meta
Pat Gunn
without running any conda commands :)
I work in academia and we're trying to figure out a backup policy that lets us skip backing up conda while still preserving the info needed for users to recreate their environment if need be. We don't want our backup scripts to su to the user to run conda stuff.
@pgunn there might be a better option but there is a list of package jsons in $HOME/.miniconda3/envs/<my_env>/conda-meta
that's assuming miniconda but i guess it's probably similar for other install methods
you could write a shell script to parse the package name out of the filename
Jonathan J. Helmus
The conda-meta directory exists for every conda environment
Pat Gunn
that'll do
Christian Roth
When I do conda create -n testfoo -c bioconda --override-channels scanpy --dry-run (which is not solvable), I do not get any Information apart from: UnsatisfiableError: - is that a regression?
Jonathan J. Helmus
Note that --override-channels means that no additional channels are searched for dependencies. Since bioconda does not provide many of the core packages, e.g. python, matplotlib no solution can be found.
Christian Roth
The solver error was expected, I just expected it to give me the same results as if I had removed the conda-forge channel from my .condarc and was a bit surprised.
Jonathan J. Helmus
Yes, older versions of conda gave more information, feel free to open an issue
Antoine Prouvost


Is the information in Compiler tools / Aside CMake SysRoot still relevant for recent version of CMake/conda?
I see the CMake issue refered to is still open, but I can build a conda package with CMake without cross-linux.cmake toolchain file without issues (except for $AR not being properly detected). Will it break once in someonelse's computer?

Sorry if this can be answered with the current doc. I'm quite new to CMake and conda-build, and this is a lot of information at once.

Wolf Vollprecht
I tried to sign up for a paid account on Anaconda.org and I get an email that tells me that my email isn't associated with an Enterprise account
Dave Hirschfeld
We made the switch today from the old repo.continuum as it was giving us 404s to https://repo.anaconda.com/pkgs
there is some strange intermittent issue yielding this:
curl: (35) OpenSSL SSL_connect: SSL_ERROR_SYSCALL in connection to repo.anaconda.com:443
it happens maybe 20% of the time when cURLing to test
downloading a package
the network admins are suspicious it is something to do with cloudflare
we are sending our CLIENT_HELLO - OUT but do not get a SERVER_HELLO back IN
Wolf Vollprecht
just wondering, i made my own channel on anaconda.org, but it doesn't have SHA256 data in the repodata? https://conda.anaconda.org/robostack/linux-64/repodata.json
is that expected or am I doing something wrong?
Hey all, I'm trying to configure a multi-user conda environment. I have a docker container that installs conda as the root user for a Jupyter notebook. At runtime, new Linux users are dynamically added. I want those users to be able to add packages to the environment, but they don't have write permissions. I've been trying to take a group-based approach, but if I chown at the end of the Dockerfile, it adds approximately ten minutes to my build time. I'd like to preconfigure the ownership so that I can add/remove packages in the Dockerfile without having to chown at the end of the process. Is there an idiomatic way to handle this kind of problem? I've been referencing https://medium.com/@pjptech/installing-anaconda-for-multiple-users-650b2a6666c6 so far.
Wolf Vollprecht
not sure if this can help you
Thanks, Wolf. Def a start.
Jonathan J. Helmus
@wolfv sha256 data is only provided for channels like conda-forge that are served from a CDN. Other channels will only have md5 hash information.
Philippe Blain

hi everyone ! I just clone the conda repo on my macOS 10.11 system and did source dev/start. I get conflicts:

Solving environment: ...working... 
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.

UnsatisfiableError: The following specifications were found to be incompatible with a past
explicit spec that is not an explicit spec in this operation (conda):

  - conda-build -> conda[version='>=4.1,<4.3|>=4.1|>=4.1,<4.4|>=4.3|>=4.5']

The following specifications were found to be incompatible with each other:
# and then pages after pages of conflicts...

I'm not sure what to do about that..

Philippe Blain

it appears my ~/.condarc file, with the following content, created that situation:

auto_activate_base: false
changeps1: false
  - conda-forge
  - defaults
channel_priority: strict
anaconda_upload: false

the super long list of conflicts ended with "Note that strict channel priority may have removed packages required for satisfiability.", which made me think of trying again after moving my ~/.condarc out of the way...

Steven C. Howell
I'm experiencing an issue where my Jupyter instance cannot activate kernels other than the one Jupyter is actually installed in. I get the error, "'activate' is not recognized as an internal or external command, operable program or batch file."
I think this is caused because the conda bin command is not actually on the path. I am on Windows 10, with the user-only install of Miniconda.
Should I file an issue or is there a simple way to correct this?
Anthony Scopatz
Is there a way to get conda to spit out the files that is actually installs, rather than what is in the info/files of the tarball
Wolf Vollprecht
I don't know, but the only special case is with noarch packages I believe
and with them, it depends on your Python version what files get installed
e.g. `pycache/....py38.pyc or something like that
Anthony Scopatz
Yeah, I think noarch is the only special case too, but there is also noarch generic, etc