I don't know anything about eglot unfortunately, but the key is make sure eglot is sending the correct project root to clojure-lsp, lsp-mode has the workspace concept, eglot should have something similar

It does send the correct project root. Problem is that it sends the same project root to multiple clojure-lsp instances. One for each of clojure-mode, clojurec-mode and clojurescript-mod. 😞

Oh, that's weird, never saw that with lsp-mode, so not sure why that is happening for eglot :(

I suspect, egot has a mapping of major mode to lsp server command. And since we are using multiple major modes for Clojure, this happens.

lsp-mode does the same, it has a mapping that triggers lsp after entering in a major mode, this is correct, but what lsp-mode does is check if that buffer/file is associated to a running workspace

I will dive into Eglot code...

You may be interested on https://clojurians.slack.com/archives/CPABC1H61/p1664137536766069

You can use consult-imenu to get a list of all functions in the current buffer and jump to them. And consult-imenu-multi for all functions for all buffers in the current project. This is just one way to do it.

Yay! Thank you.

definitively but first have to ask if something have been already done. Was looking at cider recently which detect a bb.edn file.

Maybe ask in #babashka as well

Here's a start, if you want to give it a shot yourself:

(let ((default-directory "~/dev/borkdude/blog")
      (options (s-split "\n"
                        (s-trim
                         (shell-command-to-string "bb tasks")))))
  (completing-read "which task?" options))

wow really cool !

Emacs sure is hackable 😁

You might be interested in projectile-run-shell-command-in-root to list tasks relative to the current project root, rather than in a specific directory.

Oh didnt know about projectile-run-shell-command-in-root

There is also project-compile which is similar and built-in. And both commands project-/projectile-compile have autocomplete.

when you're done with this, feel free to post the elisp snippet to the babashka wiki

Yes! 🙂

I see the list in the minibuffer in a different order… I would love to see them vertically stacked in a buffer in the order they actually appear (but maybe I’m using imenu differently).

you can use occur and look for def

Oh yes! That’s perfect!

even shows the first line and args. I should be able to create something that would show the second line of each form, then, if I have docstrings, I’d be even more set.

i think occur can take args of how much context to show

check out m-x describe-function [ret] occur

I tried this, which didn't do what I wanted:

(defun teod/tryit ()
  (interactive)
  (replace-regexp-in-region "\\.\ " ".\n"))

reading https://www.gnu.org/software/emacs/manual/html_node/elisp/Regexp-Special.html now

M-x re-builder is useful for experimenting with regex in emacs. emacs regex syntax is slightly weird.

re-builder is awesome! "\\. " might be what I want.

I ended up going a different route:

(defun teod/one-line-per-sentence ()
  (interactive)
  (let* ((s (delete-and-extract-region (mark) (point)))
         (s2 (s-replace ". " ".\n" s)))
    (insert s2)))

Not relevant for your original question but you can check if the region is actually "active" instead of using (point) and (mark). I always use something like this in my functions that have to take a region:

(defun py2json (start end)
    "Read region in Python syntax and convert to json."
    (interactive
     (if (or (use-region-p) (not transient-mark-mode))
         (prog1 (list (region-beginning) (region-end))
           (deactivate-mark))
       (user-error "No region")))
    .... 

(with-temp-buffer
  (insert "Sentence one. Sentence two. Sentence three.")
  (goto-char (point-min))
  (replace-regexp-in-region "\\." "\\&\n" (point-min) (point-max))
  (buffer-string))
;; =>
"Sentence one.
 Sentence two.
 Sentence three."

Nice, thanks!