Tooltip selector regression in v3.3.0

[rymohr]

The selector handling changes introduced in 3.3.0 break manual tooltips initialized with a selector.

http://jsfiddle.net/sk5o25b9/ (worked fine in 3.2.0)
http://jsfiddle.net/sk5o25b9/1 (busted in 3.3.0+)

See PR #14189 and related commit 1b32376
Related to #14167

[mattlunn]

Youch. The change you reference prevents any "use" of the tooltip/ popup after you've initialized it; e.g you can't pass a string to $().popover() or $().tooltip() anymore (e.g. $().popover('destroy').

A solution (!?) would be to alter the signature of Popover(option) and Tooltip(option) to be Popover([selector], option) and Tooltip([selector], option), e.g.:

$('foo').popover({
  selector: 'mark',
  trigger: 'hover',
  container: "body",
});

// later...
$('foo').popover('mark', 'destroy');

This would then give Tooltip() and Popover() functions enough context in later invocations to know which instance the user is trying to target. Obviously if you don't specify a "selector", the "selector" argument won't need to be provided in further calls, and you could get away with $('foo').popover('destroy') directly.

See also http://stackoverflow.com/q/27108979/444991

[redbmk]

I believe I have a fix for this: 465ef28

[redbmk]

Nevermind...that seemed to have fixed it for tooltip selectors but broken it for tooltips without selectors. I'll see if I can find a more complete solution tomorrow.

[redbmk]

I can't come up with a good, backwards-compatible way to fix this. I ran into this issue because I'm calling $().tooltip("destroy") where there is a selector, but @mattlunn's proposal would mean I would have to change my code to add the selector in there. It seems like it might be best to revert 1b32376 and look into a new way to handle selectors for v4.0 that fixes #14167.

[rymohr]

Only thing I can think of would be to track the last selector given, and use that as the default for any future calls.

[cvrebert]

Maybe instead of storing the instance data per-selector, we could add an explicit instanceId (or whatever) option and store the data per-ID. And when no ID is provided, default to some special ID so that stuff not using selector still works.
Just a random thought.

[cvrebert]

@hnrch02 Any objection to reverting #14189 for the time being?

[hnrch02]

@cvrebert No, seems like this is a bigger issue than #14167. We'll see if we can address it in v4.