viewport units test

[paulirish]

via @chriscoyier

basicallly 100vw == window.width

test to be written for real.

[chriscoyier]

Yay. There is a real use-case I think especially now that it's shipping in Chrome 20 & IE 10.

If support, use.
If no support, polyfill with FitText.js

[ryanseddon]

IE9 and maybe 10 interpret 1vw as the full width and not 1/100th of the width.

http://jsfiddle.net/ryanseddon/x4jA9/

[chriscoyier]

It was fixed in 10, in my testing. This is kind of a disaster though. Maybe Modernizr should just report no support for IE 9? Or would you do tweener/broken support like .viewportunits-100x

[tnorthcutt]

+1 for reporting no support in IE 9, at least initially.

[ryanseddon]

@chriscoyier yeah I reckon just say no support for IE9 since it's a broken implementation anyway. Your proposed test will fail in IE9 so that should be fine.

I modified my testcase based of your idea to test for 50vw width as it would false positive in Firefox doing 100vw as the div it injected would naturally be the full window width too.

http://jsfiddle.net/ryanseddon/x4jA9/12/

This is false in IE9 and true in Chrome 20 and IE10

[FWeinb]

@ryanseddon if window.innerWidth is odd it won't work because getComputedStyle(elem, null)['width'] is rounded down.

This might be better: http://jsfiddle.net/FWeinb/etnYC/

(Tested in Chrome 20)

[ryanseddon]

@FWeinb is anything changed I don't see any difference in the link?

[FWeinb]

@ryanseddon Sorry I didn't saved it. I wrapped the window.innerWidth/2 with parseInt() to equally round it down.

[ryanseddon]

Ok awesome, does someone want to put a pull request together?

[paulirish]

@sguolin want to take a shot at the PR for this?

[glsee]

Yeah, sure! Kindly review: #638
P/S: changed my username from @sguolin recently.

[sindresorhus]

Fixed in #638