W3C home > Mailing lists > Public > www-style@w3.org > April 2011

Re: [css3-lists] [css3-speech] Interaction between list-style-type and speak properties

From: Tab Atkins Jr. <jackalmage@gmail.com>
Date: Wed, 27 Apr 2011 13:16:16 -0700
Message-ID: <BANLkTikTd1uv-vDj6urM1Kv_75F185U28A@mail.gmail.com>
To: fantasai <fantasai.lists@inkedblade.net>
Cc: www-style@w3.org
On Wed, Apr 27, 2011 at 11:55 AM, fantasai
<fantasai.lists@inkedblade.net> wrote:
> On 04/27/2011 11:43 AM, Tab Atkins Jr. wrote:
>>
>> On Wed, Apr 27, 2011 at 11:24 AM, fantasai
>> <fantasai.lists@inkedblade.net>  wrote:
>>>
>>> On 04/27/2011 11:12 AM, Tab Atkins Jr. wrote:
>>>>
>>>> It doesn't seem like the distinction between numeric and alphabetic is
>>>> important.  They're just alternate ways of representing numbers.
>>>> We've already made the point that legal documents, where the precise
>>>> marker is important, should use inline text for their markers (which
>>>> reminds me that I need to add the
>>>> display:marker/list-style-type:inline feature).  In other documents,
>>>> the fact that a list is presented as "A" instead of "1" is mostly
>>>> irrelevant.  This is styling information, not semantic content.
>>>
>>> Yes, it's styling rather than semantic. I'm not arguing semantics.
>>> I'm arguing that if the document is styled with letters, then an
>>> aural presentation of it in all likelihood wants to read those letters
>>> rather than treating it as bullets (<ul>) or numbers (<ol>) depending
>>> on the markup. Daniel's point is that this capability is not addressed
>>> in either CSS3 Speech or CSS3 Lists.
>>
>> And my point is that even if you make the distinction between bullets,
>> numbers, and alphanumerics, there are still list styles that can't be
>> slotted into those categories.  Even within a seemingly-simple style
>> like 'alphabetic', you can get styles that *cannot* be read - for
>> example, the "go stones" example I have in the spec.
>
> Which means there's a problem that needs to be solved, not that there
> is no problem.

The problem is that you're trying to read out style information as
semantic content.  Don't do that.  This is similar to the fact that
CSS-specified images don't have alt text, and thus you can potentially
lose information when a screenreader reads the page, if you were using
those images to convey information.



On Wed, Apr 27, 2011 at 12:24 PM, Belov, Charles
<Charles.Belov@sfmta.com> wrote:
> Alas, we are seeing a side effect of comment-trimming when replying.
> This whole thread started with my post on Monday 2/7/2011 at 12:56 p.m.,
> which stated a real-world case as to why it was a problem for screen
> readers not to read the <ol> bullets as written:
>
> "in the case of a public meeting, where we have a legally published
> agenda, and items are called by the chair by letter, it would be
> important to me that the rendered speech be:
>
> A. First item
> B. Second item
> C. Third item
>
> and I would definitely *not* want to leave this decision to the user
> agent."

So, this is already up to the user-agent visually.  If the UA is
ignoring CSS, or there's a network error making the stylesheet load
fail, or some other circumstance that results in the "list-style-type"
property not being properly applied to the list, you'll lose the
letters as they flip back to plain numbers.  Similarly, if you use the
counter() function to number your headings, and the CSS is lost,
you'll lose your heading numbers.

The correct answer to this case is that, when the list markers are
vital to the content, they should be part of the content.  When
they're not vital, the loss of information from the style being lost
or not inferable is unfortunate, but not killer.  If you really need a
way to control how list markers are read out beyond the basic "bullets
vs numbers" distinction, this can be addressed by a specialized
property in Speech, but I don't personally think it's vital to
address.

~TJ
Received on Wednesday, 27 April 2011 20:17:03 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:39 GMT