I'll be refactoring some Clojure code today and adding some new namespaces, so should give the completion a nice little workout. I'll first download the latest clojure-lsp server and update all the Emacs packages (and Spacemacs config). If there are issues across Emacs configs (i.e. Doom, Spacemacs) then it suggests some feedback to the LSP projects is required, so I'll try create a project with reproducible issues.

Thanks john, i'll try to do the same. The hard part here is always it works flawlessly on my toy projects and only intermittently on the larger ones. And in the larger ones, that i can't share, i'm not sure how to report the relevant information! I think i want to learn more about emacs internals and elisp, need some sort of introduction material though. it seems more daunting then clojure thats for sure.

I have some interesting cases that i think form a pattern around when completion works and doesn't work. I'm on doom right now, but i'm guessing the same thing will happen in spacemacs. the require for [kyber.utils :as ku] doesn't have completion on ku/ but if i do [kyber.utils :as con] it does work on con/.* My guess is that con/ works because con triggers at least clojure conj and cons. That is, it finds some set of functions to complete and then finds more the aliased kyber.utils ns as con. [kyber.utils :as xt] doesnt work but [kyber.utils :as in] does, i believe, bc cljs.core has inc, indexed? infinte, etc...

i'll try this on spacemacs.

why this would be, is beyond me, but i'm guessing there is a race condition where if it can't find a completion fast enough it gives up? But if it does find one it keeps searching for more.

Magit forge warning. If you have updated packages in the last 8 days you may have lost the @ forge keybinding. There is now a https://github.com/syl20bnr/spacemacs/pull/15167 that brings the keybinding back again, so pull down the latest develop branch for Spacemacs.