I am a very lazy person and want to know if the functionality exists for this. If not, can anyone give me a quick course in the completion handling api? I was unfortunately a little too slow to grasp the big picture from the source code and slightly limited documentation.
There’s no functionality for this currently.
We could think about patching the REPL stdlib to allow fuzzy completions, but imho those don’t work well in a REPL (and most CLI autocompletions aren’t really fuzzy).
That said: the commit I linked above works fairly nicely in Juno, so maybe we should switch to a custom implementation instead of re-using the REPL.
Very cool! I can’t speak for others but I know this will definitely improve my quality of life. I will admit to having long term interest in hooking up some Flux based Neural network for suggestions so any separation would be useful!
I opened an issue about this in Juno.jl repo, and if you have any idea related to this, commenting there as well would be helpful.