Run above test case on Safari.
Actual result: failed
Expected result: pass
The same test case passes on Firefox4.6.
Also fails on both Opera 10.53 and Chrome 5.0.375.38 beta on Snow Leopard.
I'd love to know what's going on with our font metrics here. The reason we fail this and the other canvas baseline tests is down to the ascent and descent we calculate for the font before drawing it. I imagine this varies between platforms but I've tested it on Chromium Linux.
Tony - this must be FontMetrics 101, can you shed any light on why Firefox and Chromium Linux (for example) return different ascent/descent for the same font?
The person at http://mudcu.be/journal/2011/01/html5-typographic-metrics/#baselineCanvas looked into this a bit and concluded it was 'invisible space'. :)
I actually know very little about font metrics and don't know where this extra 'invisible space' comes from. Levi or Emil probably know more.
(In reply to comment #3)
> I actually know very little about font metrics and don't know where this extra 'invisible space' comes from. Levi or Emil probably know more.
I'm afraid this one is beyond my expertise as well :-/