Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

YCM prevents Jedi show_call_signatures from working #1890

Closed
oblitum opened this issue Jan 8, 2016 · 29 comments
Closed

YCM prevents Jedi show_call_signatures from working #1890

oblitum opened this issue Jan 8, 2016 · 29 comments

Comments

@oblitum
Copy link
Contributor

oblitum commented Jan 8, 2016

Jedi has an option, g:jedi#show_call_signatures, which when set to 1 or 2 (with noshowmode) will let one see function call signatures after typing (. When YCM is active this doesn't work, even when g:ycm_filetype_specific_completion_to_disable = { 'python': 1 }, I recall setting this before allowed Jedi call signatures to work.

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

On ycm-core/ycmd#292 you said that it was the start to enable #1300. If we do that then we would be able to basically reproduce g:jedi#show_call_signatures directly in YCM so IMHO we would have to see how much would cost us to work on a compatibility issue considering that then we would override this behaviour anyway. I understand that in the meantime would be really bad for users so maybe what I'm saying doesn't make much sense 😕

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

@oblitum it doesn't work for any config of g:jedi#show_call_signatures? or only for one for the two?

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

It makes sense, I'm just not sure how far 1300 for python is, I'd like to get what I was used to now, anyway, this is not the most important point about this issue, I just wished to have the problem registered.

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

@vheon both 1 and 2 don't work.

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

Just to be sure, are you using your fork or YouCompleteMe from master?

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

@vheon in truth, I'm at some random local branch for the swift completer... I'm switching to upstream master to check.

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

@oblitum I wanted to know if you were already using the JediHTTP version or the old one, thanks.

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

same behavior on master.

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

my master is at 07f4402.

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

Ok so it isn't a JediHTTP thing. To your knowledge does jedi-vim have some kind of logging?

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

well, that commit does deal with something omni func related =), not sure whether it just happen to be the one, but my previous usage of g:ycm_filetype_specific_completion_to_disable = { 'python': 1 } was to let the omni completer for Jedi (I think), iirc it worked in the past.

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

no idea about Jedi logging. I just verify that Jedi is being triggered on . because it's a bit slower than YCM.

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

what is the value of &omnifunc after you at least once have entered Insert mode?

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

@vheon jedi#completions

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

😭

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

The fault is not in that commit, I'm trying to find a commit where this has worked... nothing yet, not sure why this is not working anymore.

@vheon
Copy link
Contributor

vheon commented Jan 8, 2016

I just installed jedi-vim and after I enter Insert mode I get youcompleteme#OmniComplete as &omnifunc

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

@vheon have you set g:ycm_filetype_specific_completion_to_disable = { 'python': 1 }? that's on purpose to let jedi work.

@oblitum
Copy link
Contributor Author

oblitum commented Jan 8, 2016

For sake of information. I do the same kind of setup with neocomplete (when I use it for python 3). It doesn't interfere with jedi call signatures.

@micbou
Copy link
Collaborator

micbou commented Jan 9, 2016

Could you try with let g:jedi#show_call_signatures_delay = 0?

@oblitum
Copy link
Contributor Author

oblitum commented Jan 9, 2016

@micbou thanks! that not only fixed the issue, the signatures are showing up faster. Not sure whether I should leave this open, since this tweak doesn't solve the default setup. I'm affraid whether this conflict may be something that can touch other plugins in other situations.

@micbou
Copy link
Collaborator

micbou commented Jan 9, 2016

Yes, let's keep this open. There is definitely an issue when g:jedi#show_call_signatures_delay > 0.

@NoAnyLove
Copy link

I had the same issue months ago. I tried neocomplete, it can work work with jedi-vim without break it call signature hint, but it can't show jedi-vim and UltiSnips candidates at the same time, and its configuration is quite complex. So I switched back to YCM.

I had a comment in issue#234, mentioned about the g:jedi#show_call_signatures_delay setting. If it's not 0, the hint will only show up when you move cursor out and forth. The possible affected code is in https://github.com/davidhalter/jedi-vim/blob/master/autoload/jedi.vim, from line 413 to 423. But I didn't learn vimscript, I have no idea what it really happened.

BTW, wish the pull#1300 can directly generate jedi#show_call_signatures from ycm.

@oblitum
Copy link
Contributor Author

oblitum commented Jan 9, 2016

@NoAnyLove Thanks for the info.

@micbou
Copy link
Collaborator

micbou commented Jun 27, 2017

This seems to work with g:jedi#show_call_signatures_delay > 0 now. @oblitum Could you confirm?

@oblitum
Copy link
Contributor Author

oblitum commented Jun 27, 2017

@micbou Yes, feel free to close this, the default jedi's call signatures hack is unstable and can be incompatible with anything. This is my current Jedi setup.

@oblitum oblitum closed this as completed Jun 27, 2017
@oblitum oblitum reopened this Jun 27, 2017
@vheon
Copy link
Contributor

vheon commented Jun 28, 2017

@oblitum is there a reason why you re-open this?

@oblitum
Copy link
Contributor Author

oblitum commented Jun 28, 2017

@vheon I've closed by accident. I've ended with "feel free to close" b/c it's up to YCM to chose to be compatible with Jedi's default call signature feature or not.

@micbou
Copy link
Collaborator

micbou commented Jun 29, 2017

Let's close since it's now working (apparently, PR #2657 didn't fix the issue so it's probably a change in jedi-vim). We can always reopen if it breaks again.

@micbou micbou closed this as completed Jun 29, 2017
@github-actions github-actions bot locked as resolved and limited conversation to collaborators May 21, 2021
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants