Is there something wrong with editing the code in question in its own
dedicated buffer
where lsp-<language> is active then cutting and pasting onto Org Babel?
I mean I think
creating something like lsp-babel would involve activating multiple
language servers
simultaneously for each language in an org buffer (you can have any
number of languages
represented in blocks within an org file)...what's the overhead for that?
On 9/2/25 1:03 PM, Ângelo Lopes wrote:
Code editing features in Org Babel are limited, since packages for
completion, documentation, and navigation tend to break inside the Org
buffer. While this isn't a major issue in notebook-style workflows
(we're not building full systems there), it does restrict us to
relying solely on native major modes for each programming language.
Would it be possible to create an LSP client for Org (an "org-lsp")
that captures the contents of source blocks and sends them to a
language server? Maybe something integrated with Eglot?
I've been using Org Babel for data analysis and model training, mostly
with ob-python. Personally, the most valuable feature would be seeing
function docstrings inline — though ob-python is already quite good as
it is.
Org Babel is amazing. Thank you for the great work!
