conda-incubator / conda-store

Data science environments, for collaboration. ✨
https://conda.store
BSD 3-Clause "New" or "Revised" License
143 stars 46 forks source link

[ENH] - Distribute Conda-Store with a bundled set of environments / pull environments from a server #552

Open dharhas opened 1 year ago

dharhas commented 1 year ago

Feature description

As we consider the local/standalone experience, there is a need to distribute a set of environment to the end user.

This could be done in several ways.

a) Distribute a set of default environments with the initial installation b) Connect to a server and download available environments from the server

My preference is b, since it is more general and enables the user to get updates to environments

The way I envision this is that there would be a new grouping for server based environments

i.e. currently we have personal & shared

image

There would be a new group for environments from a server say myserver that would list environments available from that server.

There are a bunch of workflow and UI questions that this brings up. Should these environment be auto downloaded and created? Or should it be on demand? Should they auto update etc? How do we deal with platform differences etc

Value and/or benefit

It allows an organization to publish curated environments that can be used by end users on local systems.

Anything else?

No response

costrouc commented 1 year ago

As we consider the local/standalone experience, there is a need to distribute a set of environment to the end user.

I'm thinking of this issue in the same way that docker registries address this issue. There should be some mechanism to push/pull environments from one server to another.

There would be a new group for environments from a server say myserver that would list environments available from that server.

This feels sort of like federation or environments to me.

costrouc commented 1 year ago

Curious what other's thoughts are on this

trallard commented 11 months ago

Discussed this issue and agreed the best path forward is to allow conda-store users to pull environments from a remote server or the like.