identifier-based completion doesn't work and clean fast enough about emacs-ycmd HOT 4 CLOSED

Here's a screenshot for the smooth effect in Vim.

from emacs-ycmd.

in Vim, just after I type this line, immediately I can get completion when I type something like 'cap'.

That's because the YCM client looks at every char you type, notices when you finished an identifier and immediately sends the new identifier to ycmd so that it can be included in the database.

In addition, after I delete that line, it will be "cleaned" from the identifier completion candidates.

That's because you probably exited insert mode in Vim to delete the line. When you do that, YCM ships the entire buffer to ycmd for re-scanning of identifiers (which will remove identifiers that don't exist in the file anymore).

I don't know what kind of magic @Valloric did for Vim client, but it works very smoothly.

YCM has been around for years now so it has covered lots of corner cases. emacs-ycmd is still young, so you need to give it some time.

from emacs-ycmd.

Thanks for explaining.

YCM has been around for years now so it has covered lots of corner cases. emacs-ycmd is still young, so you need to give it some time

Yeah, I definitely agree. It's already making rapid progress. I just bring this up.

from emacs-ycmd.

There's definitely room to lots of improvement in the performance of the emacs client. Right now we just re-parse a file every two seconds, and that of course leads to lags. I'm going to convert this issue into separate ones that reflect @Valloric's specific techniques, but I'll reference this one for posterity.

from emacs-ycmd.