-
Notifications
You must be signed in to change notification settings - Fork 232
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Allow flexible channel priority #1254
Comments
He @dp-alvarez Do you build your custom channel on top of conda-forge? This is a known request, the latest response is: #1029 (reply in thread). With enough people wanting this, we probably need to give in at some point 😉 Track it by giving a thumbs up! 👍 |
Ooh I didn't that thread it as I only searched in issues, thanks for pointing it out. We don't build it on top of conda-forge, that channel only has our own internal packages plus patched versions of some upstream libraries. |
There is also some work ongoing in |
Oh that is interesting. Are you building everything from scratch? Including all compilers etc? |
Sadly it's nowhere that cool. We just build patched versions of libraries (pulling dependencies from conda-forge) and upload them to an internal conda channel. @wolfv great to know there is already code underway for it, thanks for the link! |
Just saw that the |
Allows the user to turn off strict priority: ```toml [project] # Turn off channel-priority = "disabled" [feature.no-default] # Turn back on channel-priority = "strict" [environments] no-default = { no-default-feature = true, features = ["no-default"] } error = ["strict"] # mixing strict and disabled errors ``` Closes #1254 #1532 #1029 --------- Co-authored-by: Bas Zalmstra <zalmstra.bas@gmail.com>
Problem description
I'm trying to get introduce Pixi at the company I work at, when I ran into this issue compared to our old conda setup.
Basically we have our own internal conda channels with custom versions of some conda packages. Some projects depend on the custom version of those libraries, some depend on the upstream ones.
Pixi, according to the docs here, follows the behavior conda has with "strict channel priority" set.
We can work around it by setting the channel source on a per-package basis, but that won't scale for wider adoption.
Is there a plan to implement flexible channel priority in Pixi? Does the solver have this behavior and it's only a matter of exposing it in Pixi?
The text was updated successfully, but these errors were encountered: