[Notebook] Deactivate tooltip on tab by default.

[Carreau]

Codepath is not removed by this.

Should I cleanup the codepath also ?

Still need to search in documentation if/where we say that tooltip is triggered by tab.

(also people will hate us for some time)

[damianavila]

Yep, they will hate us, jeje... but I think, at the end, is a good solution.
+1

[ivanov]

i missed where this came from, but it makes me 😦 to not be able to just tab to get the tool tip.

I am still able to sometimes trigger the tooltip by pressing tab a bunch of times. The completions I get with this branch don't make sense, either (why should cell magics pop-up first as arguments to a function call?)

[damianavila]

@ivanov this came from the last meeting... there was a discussion about this and the general agreement was to separate completion from the tool tip...

Btw, when I tested, I did not see the tooltip... even pressing the tab a bunch of times....

But, testing again, now I see the "cell magics" completion... I did not see it before because I was completing over a letter that did not match any cell magic...

[Carreau]

I missed where this came from, but it makes me :-( to not be able to just tab to get the tool tip.

The rationale (on a muti language env) is that to be able to decide whether to use oinfo or complete there have to be some lexing in the frontend if both are bound to tab. So either we extend the message spec (to decide in kernel), or we split help/complete action on 2 keys.
Shift tab exist for quite some time now and was already configurable, shift tab have also the advantage of working on selection in the middle of code.

I personally was for the first solution of asking the kernel, but in some case there is no way to decide what the user wish.

I am still able to sometimes trigger the tooltip by pressing tab a bunch of times.

Will try but I don't understand why...

The completions I get with this branch don't make sense, either (why should cell magics pop-up first as arguments to a function call?)

This is how completion behave in all frontend, not notebook specific :

In [3]: histogram(
Display all 343 possibilities? (y or n)
%%!                        %loadpy                    %who                       UnboundLocalError          contrib                    long
%%HTML                     %logoff                    %who_ls                    UnicodeDecodeError         copyright                  map
%%SVG                      %logon                     %whos                      UnicodeEncodeError         credits                    max
%%bash                     %logstart                  %xdel                      UnicodeError               def                        memoryview
%%capture                  %logstate                  %xmode                     UnicodeTranslateError

[ivanov]

thanks @Carreau - i understand better now why this needs to happen. I suppose it's a little too much work to just weaponize Tab and let the kernels dictate the behavior that key should implement ...

[ellisonbg]

I am +1 on cleaning up the code.

[Carreau]

As per @ellisonbg request, logic cleaned. Rebased on master also.

[rossant]

Excuse me, but how do I get the tooltip back in the notebook? Tab does not work anymore (move focus to next cell), shift+tab moves focus to previous cell.

[Carreau]

Can you try to force refresh see if this is not stale cache ? Will retry this branch more extensive though.

[rossant]

@Carreau no luck -- it's been a few days. Same problem in latest Firefox, Chrome, Opera, IE11.

[Carreau]

with this branch or on master?

[rossant]

ah sorry, master. Can you reproduce this problem?

[Carreau]

does the JS console says anything ?

[rossant]

[13:17:25,343] TypeError: cell.kernel.object_info_request is not a function @ http://127.0.0.1:8888/static/notebook/js/tooltip.js:228

[Carreau]

Ok found the reason. something that comms broke.

[rossant]

Ok, thanks for your help! I let you take it from here :)

[minrk]

Anything else here?

[Carreau]

I don't see anything else to add remove..