Fix filtering from visualizations

[Bargs]

The ability to create filters by clicking on visualizations was broken
between 5.0 alpha3 and alpha4. The root cause was the addition of a new
parameter being sent to all visualization event handlers, which was
added to support persistence of tilemap position and zoom level. The
filter_bar_click_handler (which enables the vis filtering behavior)
already expected a different kind of second parameter, so the addition
of this new parameter threw off the click handler's internal logic.

It turns out that the new parameter is unnecessary, because the object
it's passing is already available on the event object. So to fix this,
I removed the new parameter from the emitter and updated the tilemap
handlers to grab the object off of the event.

Fixes: #7501
Related: #6835 (added the new param)

[Bargs]

@tbragin perhaps you'd like to test this one as well?

[jbudz]

Awesome, testing now and everything is working. can you point to where the third argument was breaking things, it's not immediately clear to me.

[jbudz]

Is it worth adding a test for not passing three arguments?

[Bargs]

@jbudz

kibana/src/ui/public/filter_bar/filter_bar_click_handler.js

Line 61 in cd76078

if (!simulate) {

When simulate is truthy, the filters don't get added to the app state. I didn't look into where simulate actually gets used, I assume it's for testing.

[Bargs]

I think a test is worthwhile, lemme try to whip something up.

[Bargs]

@jbudz pushed a test

[jbudz]

LGTM

[Bargs]

@ycombinator since you're familiar with the PR that introduced the bug, would you mind giving this a second set of eyes?

[tbragin]

@Bargs I did a quick test and works for me functionally. Thank you so much for your help on this!

[ycombinator]

Functionality LGTM. Reviewing code now...

[ycombinator]

The root cause was the addition of a new parameter being sent to all visualization event handlers, which was added to support persistence of tilemap position and zoom level. The filter_bar_click_handler (which enables the vis filtering behavior) already expected a different kind of second parameter, so the addition of this new parameter threw off the click handler's internal logic.

Thanks for this clear explanation. Can we have type safety now? 👅

[ycombinator]

LGTM. 👍 for adding that strict unit test.