Ben Jackson

Interesting, I can't repro in my version of vim, which is only a few patches behind yours. ``` (env) ben@BeniMac2020 10 % vim --version VIM - Vi IMproved 8.2 (2019...

I can't repro on linux at work either.... :/

We'll need to get a stack trace of vim while the sleep is running and maybe set a breakpoint on `got_int` internally. We're not doing anything crazy in YCM but...

I still can't reproduce this with or without an extra conf file on any version of vim or any OS! I therefore don't see any hope of getting to a...

I thought using python in the tests would lead to flakiness. Have to think of something else for that.

both solutions use busy-waiting. wouldn't sleeping in select()/poll()/etc. waiting for input be better?

What specific attack vector are you concerned about ? I don't believe that 1.2.17 is easily exposed to the log4shell vulnerability, though it may well be riddled with other issues....

actually I just noticed they removed log4j recently: https://github.com/eclipse/eclipse.jdt.ls/commit/d28e26ef5d618fd00e5f527ce128ba6369353879 and there's a 1.7.0 release that contains it. I think we can try to update eclipse.jdt.ls in ycm. Standby.

Unfortunately,, the 1.7.0 version has a _massive_ regression in signature help that breaks our tests. However, I think that jdt.ls doesn't actually use log4l so perhaps it's not really affected....