https://github.com/MacPython/gensim-wheels is using the rackspace account for the release process and on top of that some gensim maintainers have dedicated gensim containers on our shared rackspace cloud storage infrastructure.
However the free hosting offer for open source projects has ended:
More details: https://mail.python.org/pipermail/scipy-dev/2020-February/023990.html
So we need to move on to something else. Please delete the files you own and no longer need in the rackspace admin interface and let's join on https://github.com/matthew-brett/multibuild/issues/304 to discuss how to setup a new release infrastructure.
Ok, I will deal with the gensim files tomorrow
Thanks.
@piskvorky I think we can safely delete all those old wheels. We don't need them anymore, because those releases have been made, and are available via PyPI. @menshikh-iv agrees that it's safe to delete them.
Shall I proceed? Or would you like a backup of these things somewhere?
I have no idea what these files are, or why they're there. Can I leave this with you?
Sure.
@ogrisel I deleted gensim-wheels and gensim-windows-wheels containers, please let me know if I need to do anything further.
Thank you very much. I am currently configuring a replacement container using anaconda.org for nightly builds for the scipy ecosystem. I will first configure it for scikit-learn and once it works I will let you know about the details.
If you want to use anaconda.org for release staging, you can update end of your scripts to upload to anaconda.org instead of rackspace. Here is the configuration I used for scikit-learn (using Azure Pipelines instead of travis and appveyor):
The important line is:
anaconda -t $TOKEN upload --force -u $ANACONDA_ORG wheelhouse/*.whl
the resulting wheels will be available at:
https://pypi.anaconda.org/$ANACONDA_ORG/simple
You can create your own anaconda organization for staging gensim wheels (e.g. gensim-wheels-staging) and create a token in the Settings / Access menu (URL should be: https://anaconda.org/gensim-wheels-staging/settings/access). You will need the following 2 permissions:
If you want to also push your nightly build wheels to the shared scipy-wheels-nightly organization, please give me the identifier of your anaconda account and I can grant you some permissions or generate a token for you.
If you do not have conda already installed, you can use pip to install the anaconda-client that provides the upload command line interface. However the version on PyPI is too old. You should rather install it from the github tag instead:
pip install -q git+https://github.com/Anaconda-Platform/[email protected]
Ok, thank you. I'm a little busy with other stuff right now, but will look into this later.
@ogrisel We ended up using S3 for staging our wheels. I'm closing this issue as there is nothing else left to do.
@piskvorky Tagging you just FYI.