by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Vinicius D Cerutti
    @viniciusdc
    Thanks @CJ-Wright
    Do you know where I can find the code that calls the conda_forge_config.yaml file ?
    Christopher J. Wright
    @CJ-Wright
    what do you mean calls the file?
    Vinicius D Cerutti
    @viniciusdc
    sorry, I want to inspect the version_update_regex and version_update_string keys
    but can't find a good way to do that, @scopatz said that you could know if we have some code from bot that reads it
    Christopher J. Wright
    @CJ-Wright
    it is loaded into the graph under the conda_forge.yml key
    Vinicius D Cerutti
    @viniciusdc
    oh ok thank you very much :smile:
    Christopher J. Wright
    @CJ-Wright
    are you waiting on my review for the profile things?
    2 replies
    Christopher J. Wright
    @CJ-Wright
    ok
    1 reply
    Vinicius D Cerutti
    @viniciusdc
    @CJ-Wright @beckermr do we always had this update prs json job on Circle ?
    Matthew R Becker
    @beckermr
    yes
    1 reply
    Vinicius D Cerutti
    @viniciusdc
    Do we have any external code for the bot ? something like conda-forge bot --admin rm-node --payload 'conda-forge::git.json'
    Matthew R Becker
    @beckermr
    let’s add this to the agenda
    Vinicius D Cerutti
    @viniciusdc
    :+1:
    Anton Akhmerov
    @akhmerov

    Hi everyone!

    For historical reasons, I have a feedstock that has an underscore instead of a dash in the package name (jupyter_sphinx, whereas the pypi package is jupyter-sphinx). That is often confusing to the users, and in particular because of that another feedstock was created with a dash: https://github.com/conda-forge/jupyter_sphinx-feedstock .

    I therefore have two questions:

    • Is it possible to delete the newer duplicate feedstock :pray: ?
    • Is there a better way to deal with this naming problem, by e.g. creating some sort of alias?
    Matthew R Becker
    @beckermr
    the best thing would be to archive the feedstock you don’t want
    TBH i’d move to the newer one with the dash
    to match pypi
    but this is best worked out amongst the two maintenance teams
    once you all know what you want to do
    let core know and we can do it
    we cannot archive one of the feedstocks without agreement between all parties
    Anton Akhmerov
    @akhmerov
    There's only one maintenance team, but I'd like to avoid user disruption.
    (I'm the maintenance team)
    Matthew R Becker
    @beckermr
    I see more than you on the new feedstock
    raise an issue there and bump the other person
    if they do not respond in a week or so, we can archive
    we won’t remove packages
    but we can prevent anymore from being made
    Anton Akhmerov
    @akhmerov
    https://github.com/jupyter/jupyter-sphinx/issues/148#issuecomment-673754677
    The other person already agreed.
    TBH I'm not sure which one to archive.
    Matthew R Becker
    @beckermr
    ahhh
    good enough for me
    Whatever you want
    you can always insert a shim package in the old feedstock that is named with the dash and installs with the underscore
    if you are worried
    Anton Akhmerov
    @akhmerov
    They both install the same thing.
    Matthew R Becker
    @beckermr
    the conda package name
    not the import name
    Anton Akhmerov
    @akhmerov
    Ah, how does that work? That'd be optimal
    Matthew R Becker
    @beckermr
    you’ll need to add outputs
    see the conda build documentation
    feel free to bump me on github if you are having trouble
    Anton Akhmerov
    @akhmerov
    I still don't understand how would it work for the user: would conda install jupyter_sphinx install the jupyter-sphinx conda package then?
    Or would it install both?
    Matthew R Becker
    @beckermr
    Both would be there, but one would just be dummy package that installs the other and doesn’t have any software in it.
    Anton Akhmerov
    @akhmerov
    I see now. It seems that this is indeed the best way forward. I should be able to figure things out. Is there any way to signal deprecation of a conda package?
    Matthew R Becker
    @beckermr
    Not that I know of
    Anton Akhmerov
    @akhmerov
    OK. Thanks for the help!