tooltip.js: getBoundingClientRect method existence check is unnecessary?

[cvrebert]

Element.getBoundingClientRect() seems extremely well-supported:
https://developer.mozilla.org/en-US/docs/Web/API/Element.getBoundingClientRect#Browser_compatibility
So then, is there a reason that there's a check in tooltip.js for whether this method exists?:

bootstrap/js/tooltip.js

Line 324 in f9ffb49

var elRect = typeof el.getBoundingClientRect == 'function' ? el.getBoundingClientRect() : null

To: @fat
X-Ref: #14090

[fat]

i feel like that was there for svg elements or something weird

[cvrebert]

Might've been due to iOS v3 not supporting it?: thebird/Swipe#71

[cvrebert]

Running Sauce cross-browser test with the existence check removed:
https://travis-ci.org/twbs/bootstrap/jobs/32660379

[hnrch02]

@cvrebert Sauce timed out, mind rerunning?

[fat]

ah yeah that might have been it. feel free to remove if tests pass

[cvrebert]

Sauce tests still pass with the check removed (see above).

[gsmet]

Hi,

I am the one who introduced this test: see #6737 : there are unrelated discussions in the PR but the point is it won't work with IE8 if you don't have this test.

IE8 does have getBoundingClientRect but it's broken, it's just an empty placeholder.

It would be better to revert this IMHO.

[gsmet]

...and add the function test I added in openwide-java@b7764d8 .

el.getBoundingClientRect is defined but is not a function in IE8.

[cvrebert]

So what we really need to do is backtrace to see when the is-a-function test was changed to a simple truthiness test, and why.

[cvrebert]

Hmm, looks like it was part of @hnrch02's changes in #14090.

[cvrebert]

Whatever the root IE8 quirk here is, it's in dire need of being documented on the MDN page (https://developer.mozilla.org/en-US/docs/Web/API/Element.getBoundingClientRect).

[gsmet]

I can assure you the code didn't work in 2.3.0 and we fixed the behavior by adding the test. Calling getBoundingClientRect returned an empty object which didn't contain anything useful.

As you can see in zeroclipboard/zeroclipboard@be45735 (it's referenced by @hnrch02 in the PR) , they removed the test BUT they added a workaround as the "method" might return something empty. This was our case and that's why we fixed it like that.

Considering the way tooltip was coded at the time, adding the function test was the right thing to do as it triggers the workaround.

[hnrch02]

I mentioned why I changed it to a simple truthiness check here: #14090 (comment). It's documented on the MDN page for the typeof operator: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/typeof#IE_host_objects_are_objects_not_functions

[hnrch02]

Also, calling getBoundingClientRect doesn't return an empty object for me:

[cvrebert]

I've documented the "IE8 getBoundingClientRect() result lacks width/height" incompatibility in MDN based on zeroclipboard/zeroclipboard@be45735 and as confirmed by @hnrch02's screencast.

[cvrebert]

Okay, I propose we fix this by doing the same thing as ZeroClipboard: #14492

[fat]

sgtm