Load repodata from historically used channels (as stored in PrefixData)
#52
Conversation
| See https://github.com/conda/conda/issues/11790 | ||
| """ | ||
| with make_temp_env("conda-forge::xz", "python") as prefix: | ||
| output = check_output( |
There was a problem hiding this comment.
Note we need a whole subprocess here because libmamba prints to STDERR directly, and that won't be intercepted with our naïve in-process capturing mechanisms.
PrefixData)
| # skip if the channel has authentication info, because | ||
| # it might cause issues with expired tokens and what not | ||
| continue | ||
| if record.channel.name in ("@", "<develop>", "pypi"): |
There was a problem hiding this comment.
Do you know if we have a definite list of these pseudo channel names in conda that we don't have to hard-code it here?
There was a problem hiding this comment.
I checked but I couldn't find it. Looks like they were added as needed for edge cases found during the development of orthogonal features. @ is for virtual channels, <develop> seems to come from pip install -e and pypi is found for pip install that do not have a conda-meta entry, as far as I can understand.
|
@jezdez Should we port this over to |
Closes conda/conda#11790
I am not sure whether we should go this way, though. I think it's a good idea, but we also need to enable this on
conda/conda,mambaandmicromambafor better compatibility across the ecosystem.Authenticated channels are the only blocker here, but it can be either worked around or handled gracefully with good errors.