ruff server with optional TCP/UNIX socket

[morph027]

I'm running pylsp --tcp as a service which listens for connections from multiple editor instances instead of creating a new pylsp process for every instance which is quite useful.

Switching to the new implementation of ruff server instead of ruff-lsp, i think, it would be nice to have such a feature too. It obviously needs to handle multiple clients.

[dhruvmanila]

What's the use case for this? Is it just to avoid creating a new ruff server process for every editor instance? What kind of benefits does that provide? This might be basic questions but I'm just trying to understand the motivation behind this feature request.

[morph027]

Jip, basically. In my previous setup, i used one pylsp instance, which serves multiple editors e.g. in tilix/tmux/... split windows and therefore saves some resources (not that much i guess).

I could however further improve my editor skills by using splits inside vim/helix which does the same :) So it's not a show stopper at all. Just left it here in case other people think the same.

[dhruvmanila]

Are you opening the same project in multiple editor instances? I think that's the only place where this could benefit. If so, I'd recommend using splits inside the editor itself :)

[bread-mountain4]

aside from the multiple editors consideration, serving the LSP over either TCP or UDS would help users whose local development is primarily done through docker. so the lsp could be running inside of the container with a local mount for the codebase and the editor could then communicate with the dockerized ruff server via tcp.