W3C home > Mailing lists > Public > public-css-testsuite@w3.org > March 2011

Re: [RC6] font-family-valid-characters-001 is not clear

From: Etan Wexler <ewexler@stickdog.com>
Date: Sun, 27 Mar 2011 16:03:26 -0400
Message-ID: <AANLkTinwpHmpaPab5kWrANjug5KAcZMEvBLHQHp3YZAK@mail.gmail.com>
To: public-css-testsuite@w3.org
Gérard Talbot (in
<http://www.w3.org/mid/99a3bdcde9c5bb49d38ae8e40202fe32.squirrel@cp3.shieldhost.com>):

> http://test.csswg.org/suites/css2.1/20110323/html4/font-family-valid-characters-001.htm
>
> looks incorrect to me.

It looks correct to me.

> Currently, all the browsers (IE8, Firefox 4, Firefox 3.6.16, Opera
> 11.01, Safari 5.0.4, Chrome 10.0.648.204, Konqueror 4.6.1) pass this
> test...
>
> First, they all put the testfoo names
> {test-foo, test_foo, test\foo, test, foo, _testfoo, -testfoo, test -foo,
> test _foo, etc.}
> between quotes, then try to fetch the font and since there is no font
> installed with such name, then move to use the alternate one Ahem.

I’m unsure of what you mean by “put the testfoo names between quotes”
(though I’m confident that “quotes” is to mean a pair of U+0022 [the
ambiguous so‐called quotation mark of ASCII fame] or a pair of U+0027
[the ambiguous so‐called apostrophe of ASCII fame]).

> Second, some font names are perfectly correct and acceptable to begin
> with: _testfoo, test\foo, test, foo and even test-foo are not invalid
> and not reported as invalid by the CSS validator. In fact, none of the 9
> declarations are reported as invalid.

All of the family names are valid in CSS and, so, all of the
declarations are valid CSS.

> Third, if the names are supposed to be invalid, then the whole
> declaration should be invalid. That's what the spec is saying:
>
> http://www.w3.org/TR/CSS21/fonts.html#font-family-prop
>
> http://www.w3.org/Style/css2-updates/draft-PR-CSS21-201103XX/fonts.html#font-family-prop
>
> So, Ahem font should not be fetched either.

Yes, if an implementation treats a family name as invalid, then the
implementation, unless fundamentally defective, would treat as invalid
the entire declaration in which the name appears. Given that each
family name is valid, implementations that pass this test accept the
dummy names as valid and, for each dummy name, seek a font family of
that name, fail to find a matching family, and fall back to Ahem, thus
obscuring the word “FAIL” by using Ahem’s nonsense glyphs.

> Fourth, when I validate the 9 rules in a stylesheet [... ,]
> the CSS validator reports no parsing error [...]

Well, kudos to the contributors to the CSS Validator. That said, the
CSS Validator is quite buggy and, therefore, is not a measure of
conformance.

-- 
Don’t include me as a recipient of public replies. I read public
replies from the list.
Received on Sunday, 27 March 2011 20:03:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 27 March 2011 20:04:08 GMT