W3C home > Mailing lists > Public > public-css-testsuite@w3.org > December 2010

Re: [RC4] bidi-alt-001 Re: RC4 Invalid tests

From: Øyvind Stenhaug <oyvinds@opera.com>
Date: Tue, 21 Dec 2010 18:08:13 +0100
To: Gérard Talbot <css21testsuite@gtalbot.org>
Cc: "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>
Message-ID: <op.vn2pvzfrbunlto@oyvinds-desktop>
On Sat, 18 Dec 2010 18:52:44 +0100, Gérard Talbot  
<css21testsuite@gtalbot.org> wrote:

>> (It was
>> fixed by fantasai by setting width:0 for img so that the alt text won't
>> be
>> visible if the img is still considered to be a replaced element.)
>
> There are browsers which will ignore, will override by default the
> author specified image placeholder dimensions (width="0", 'width: 0')
> when rendering the alt text for accessibility purposes. This is what
> happens in this testcase with Firefox 3.6.13.

The idea is that there are two possibilities:
A) An image that represents its alt text is still treated as a replaced  
element
This means
- A single box is rendered ("atomic inline-level box")
- The 'width' property applies
- The contents of the IMG (i.e. the alt text) are outside the scope of CSS
B) An image that represents its alt text is treated as a non-replaced  
element
This means
- Multiple boxes might be rendered
- The 'width' property does not apply
- The contents are within the scope of CSS

Looks like Firefox is trying to do B (long alt texts are broken into  
multiple boxes, line-height has an effect etc). Kind of like  
img{content:attr[alt];}. So there seems to be no reason  
direction/unicode-bidi shouldn't apply.

-- 
Øyvind Stenhaug
Core Norway, Opera Software ASA
Received on Tuesday, 21 December 2010 17:06:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 21 December 2010 17:06:57 GMT