Re: Browser zooms and testsuites

To respond shortly, the tests written on that day will be reviewed by CSSWG members and will be submitted at a later date.
 
--------
 
Zoom is actually a browser-trick to render on 1000px a page like if the viewport was 1250px wide, but without actually drawing a 1250px image. Except some rounding/aliazing issues, zoom should actually be a no-op.
 
So, the real issue is not that the tests *suppose* that the zoom is 100%, but it's that the browser code makes that implicit assumption, which is false. That means that websites relying on the tested features will render correctly on a "by-default" desktop browser but incorrectly for people using the zoom for accessibility reaons (or any other reasons).
 
The whole point is that browser vendors should, once in a while, test their browser with "zoom: 125%" enabled, just to catch zoom-related bugs.
 
--------
 
The fact is I *know* for sure some tests fails under zoom on IE/Chrome because someone else reported those bugs already (webkit issue had been forgotten and was reactivated, didn't check the latest status) and I posted a bug report for IE. However, I didn't take the time to test myself every test I wrote with zoom enabled. Tests are meant to be automated, so let's automate this ;-)
 
Actually, viewport units are more sensible that some other features to zoom, so we can maybe limitate that to cases where viewport units are involved.
 
--------
 
Regarding my mail address, I considered using another one many times already, but my mail system is already so complex... You'll respond it's not your problem and you'll be right. So, if you get issues receiving my mails, I can actually update change my W3C mail address to something else. But then I've to figure out which one. 

Received on Sunday, 28 October 2012 17:13:55 UTC