Fork me on GitHub

It certainly worked with nrepl:// at some point


@tpope having played with the asyncish omnicomplete on-type, I'm still noticing the delay. I am in ClojureScript where the clojurescript runs on a remote machine (don't ask - corporate envs 😞 ). So I'm probably feeling it more than others would. I guess the sleep loop means that it still blocks, but you'll see some results at a time until it is completely done? I might try creating a forked cider-nrepl to repo on the JVM where completions have some delay in them. One option might be to have a #complete which provides a callback interface. Then I can hook that up into quite easily. For now I have a working asyncomplete thing which works reasonably well with some copy-paste.


complete_check() supposedly also allows you to continue typing


i think a callback version is a good idea. maybe bolt it on to #omnicomplete if the method signature doesn't get too convoluted


Hmm, you're right. I missed that. I guess I need to debug what the cause is deeper.


@dominicm is firing off a callback once for each result sufficient? if it needs all the results at once i don't think it would be sufficient


> Rather than writing your own completion source from scratch you could also suggests other plugin authors to provide a async completion api that works for asyncomplete.vim or any other async autocomplete libraries without taking a dependency on asyncomplete.vim. I am totally on board for this in theory but they jump from "without taking a dependency" to an example with hard coded references to asyncomplete#