RE: New tests for web fonts

I've been significantly reworking the webfonts test pages, and will try to publish formally with a results page soon.  You can see the new version at the same place:

http://www.w3.org/International/tests/test-webfonts-1

http://www.w3.org/International/tests/test-webfonts-2

I also retested, adding Chrome to the list, and came up with the following results:  (all tests on Windows XP using Uniscribe v1.626.5756.0 and only latest official versions reported here, except for IE8)

Firefox 3.0.1: 
font-linking: no support
EOT: no support
Installed font text: perfect result

Google Chrome beta: 
font-linking: no support
EOT: no support
Installed font text: perfect result (cf. Safari, since Chrome is also based on WebKit)


Safari 3.1.2: 
font-linking: Georgian and Armenian, supports fine (though Georgian text looks a little bolder); Khmer, blank space; Hindi, no support, and a difference between the result for the downloaded font (blank space) and the system default (correctly rendered text); Arabic, same as Hindi; Urdu, blank space; Thai, Tibetan & Myanmar, blank space (though system default shows fonts)
EOT: no support
Installed font text: Georgian, Armenian, Hindi, Thai, Tibetan & Myanmar perfect; Khmer, a slightly higher placement of the vowel sign diacritic above the base character when using the Battambang font (not the Mool font), but otherwise fine; Arabic, in Arabic Typesetting font only, misaligned contextual join between first two characters of penultimate word, and in both fonts diacritics slightly too high, though correctly placed relative to each other; Urdu, blank space, though system default font looks ok; 


Opera 9.51: 
font-linking: no support
EOT: no support
Installed font text: perfect result


IE7:
font-linking: no support
EOT: Georgian, Armenian, Hindi, Thai, Tibetan & Myanmar perfect; Khmer, fine except that KHMER VOWEL SIGN AA, KHMER VOWEL SIGN OO and KHMER SIGN YUUKALEAPINTU are completely missing in the Battambang font and rendered as square boxes in the Mool font (note that these are spacing combining marks); Arabic, many characters are missing, and the two fonts yield slightly different results; Urdu, only alephs are produced, all other characters missing; 
Installed font text: perfect result


IE8 beta2:
font-linking: no support
EOT: no support

Once again, if you are interested in sending me results of tests you have run yourselves, please  send me screen snaps and the following info: OS, UA version, Uniscribe/Pango/ATSUI/etc version.


Things I noted in particular:
Chrome doesn't support font-linking, unlike Safari, though Safari only manages to render the text doesn't contain OT features. On the other hand, Safari produces the installed font text mostly ok (with the marked exception of Urdu), so this is down to the downloading. Unlike Safari, Chrome reproduces the installed font text perfectly. 
IE7 has  problem with spacing combining marks in Khmer, but not Hindi, and makes a complete mess of any Arabic-based script.
IE8 beta2 still doesn't support EOT :(


Again, please point out to me any silly mistakes, or serious ones, before I go public with this. Thanks.

RI

============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/



> -----Original Message-----
> From: www-international-request@w3.org [mailto:www-international-
> request@w3.org] On Behalf Of Richard Ishida
> Sent: 25 August 2008 11:43
> To: www-international@w3.org; www-style@w3.org
> Subject: New tests for web fonts
> 
> 
> I've been developing a set a of tests for downloaded opentype font and .eot font
> support in browsers for languages that use 'complex scripts', since that is a hugely
> valuable use case for web fonts across a large proportion of the world.  Lots of
> people are producing free embeddable and editable open type fonts so that
> people in their region can read Unicode based web pages. (A large proportion of
> the over 800 fonts on my system fall into this category.)
> 
> I'd be happy for people to poke around at these tests, and check that there are no
> errors, before I announce them officially.  You can find them at:
> 
> http://www.w3.org/International/tests/test-webfonts-1
> 
> http://www.w3.org/International/tests/test-webfonts-2
> 
> 
> 
> 
> Preliminary results from a quick initial test show the following:  (all tests on
> Windows XP using Uniscribe v1.626.5756.0)
> 
> Truetype/Opentype downloads
> ========================
> 
> IE7: No support
> 
> IE8: No support (and an unexpected problem displaying the Thai, Tibetan,
> Myanmar page at all)
> 
> Firefox 3: No support
> 
> Opera 9.51: No support
> 
> Safari 3.1.2: Armenian, perfect; all others yield blank space (although the reference
> text is rendered correctly for Hindi, Arabic and Urdu in the default browser font).
> 
> 
> 
> .eot downloads
> ============
> 
> IE7: Armenian supported fine; Khmer, mostly ok but drops or displays as boxes
> certain characters; Hindi, perfect; Arabic, only shows certain characters, doesn't
> place multiple diacritics properly; Urdu, only shows aleph; Tibetan/Thai/Myanmar,
> Tibetan and Myanmar are perfect, but Thai represents some combining characters
> as square boxes.
> 
> IE8: (surprisingly) No support (and a problem displaying the Thai, Tibetan,
> Myanmar page at all)
> 
> Firefox 3: No support
> 
> Opera 9.51: No support
> 
> Safari 3.1.2: No support (but behavior varies: Armenian, default font; Khmer, blank
> space; Hindi, Arabic & Urdu, default font; Thai, default font, but Tibetan and
> Myanmar, blank space)
> 
> 
> 
> 
> If you are interested in sending me results of tests you have run yourselves, please
> send me screen snaps and the following info:
> OS, UA version, Uniscribe/Pango/Core Text/etc version.
> 
> 
> Cheers,
> RI
> 
> 
> 
> ============
> Richard Ishida
> Internationalization Lead
> W3C (World Wide Web Consortium)
> 
> http://www.w3.org/International/
> http://rishida.net/
> 
>

Received on Sunday, 7 September 2008 18:58:54 UTC