Calculate offset between html and body elements

[JamesMGreene]

From a commit comment, @sporkmonger suggested we factor in the use of document.body.getBoundingClientRect() and document.documentElement.getBoundingClientRect() to calculate the top/left offsets between the html and body elements, if any, within the _getDOMObjectPosition internal function:

Should probably incorporate document.body.getBoundingClientRect().top - document.body.parentElement.getBoundingClientRect().top into the calculation here. For sites with position: fixed headers that shift the whole body element down, the offsets seem to end up being wrong for some reason. In those cases, delta between document.body.getBoundingClientRect() and document.body.parentElement.getBoundingClientRect().top has the real top offset.

[JamesMGreene]

May also be related to Issue #423.

[sporkmonger]

FWIW, I can confirm that we implemented the change on our site and it works as expected now. I just don't know if the change would have side-effects on other page layouts or not.

[vladshcherbin]

If body has margins and position:relative (for absolute items in it) - swf has wrong position.

Example

@JamesMGreene, is there any possible quick fix ?

[sporkmonger]

@vladshcherbin The code in that fiddle is minified, but from glancing at it, it appears that it doesn't include the fix I've been referring to. Try adding the fix in and see if it resolves.

[vladshcherbin]

@sporkmonger the code is minified 2.1.6, without any fixes
here is this example with 2.1.6, but not minified - link
can you please check, how it works and maybe post a fix or a link to an example with your fix ?

[sporkmonger]

https://github.com/bitpesa/zeroclipboard-rails/commit/63c6ec3034d5352c45073b1ec025c6af7ca63adf

We're using zeroclipboard/zeroclipboard-rails, so we forked that and implemented our fix there even though, if it works, the code should ultimately live in the upstream project.

[vladshcherbin]

@sporkmonger thanks, it finally works.
Confirmed on the latest version here (2.1.6)

[sporkmonger]

OK, great, independent confirmation! Anybody want to chime in and confirm that the code doesn't break sites that were already working? That's actually the detail I'm more worried about.

[JamesMGreene]

Anybody want to chime in and confirm that the code doesn't break sites that were already working? That's actually the detail I'm more worried about.

Ditto to that worry! 😛

[JamesMGreene]

Related: #495 ("Create positioning unit tests")

[robdecker]

@sporkmonger Your fix works great on version 2.1.6. Thanks!

[JamesMGreene]

Yeah, I'll try to get this and a few other fixes rolled into a v2.1.7/v2.2.0 release soon.