Re: FYI: HTML for Mobile Access

Hi,

I'm an editor of "HTML 4.0 Guidelines for Mobile Access".  Thank you
for paying attention to this Note.

Harvey Bingham <hbingham@ACM.org> wrote:

> I find seven suggestions for user agents, addressing the mobile 
> browsers. Interesting to get the set of issues particularly
> from that viewpoint. Lots of commonality. They do reference our work.

Yes, we find WAI Guidelines useful, and we do reference User Agent
Accessibility Guidelines.  Our references to UA Guidelines are based
on WD-WAI-USERAGENT-19981112, so we are checking WAI-USERAGENT-19990331
now and will try to update references appropriately.  Our basic idea
is to make our "mobile guidelines" compatible with WAI Guidelines
whenever appropriate.  When WAI Guidelines become mature, we will
simply refer to relevant sections of WAI Guidelines and will focus
on differences (if any).

> I prefix below my comments with "HB".

I prefix my comments with "MI". In my comments, "we" means members
of Mobile Access IG or at least editors of the mobile guidelines.

> 3.1. Styles
> Note for User Agents:
> It is desirable that even if user agents don't support style sheets, it
> should parse STYLE elements not to show styles nor to cause any other
> side-effects.
> 
> HB: Some style info can add prefix text. That is lost.

MI: Yes, for example, `content' property in CSS2 allows you to
generate content with `:before' and `:after' pseudo-elements [1].
We realize that, but we can't ask too much because of hardware
restrictions of mobile devices [2].

Unfortunately, some older user agents show contents of SYTLE elements
as garbage.  That's too bad, as typically screens are very small in
mobile devices, so such garbage can easily fill the whole screen.
That's worse than not rendering some contents supposed to be
generated by style info.  That's why we advise to hide the contents
of STYLE elements when user agents don't support style sheets.  This
is in line with the minimum requirement in HTML 3.2 [3].

We understand that generated content is useful for accessibility, but
basically, we think the important information should be described in
the document itself, and should not be generated by style info.
Generated content should be used to "improve" accessibility.  If
the document is not understandable without such generated content,
we think it's the "abuse" of style sheets.  So we state as follows
in section 3.1 [4].

 "Contents should be readable without style sheets so that devices which 
  don't support style sheets can still render contents reasonably."

In the future, we might use better content negotiation mechanism,
such as CC/PP [5], but it's out of scope of this Note. "Modularization
of XHTML" [6] will address this issue.

> 3.2. Scripts, Event Handlers
> Note for User Agents:
> It is desirable that even if user agents don't support scripts, it should
> parse SCRIPT elements not to show scripts nor to cause any other 
> side-effects.
> 
> HB: Doesn't that make the effect of the script NULL, losing information?

MI: The same reason for styles.  We can't ask too much to user agents.
Content developers should check "2.13 Scripts" of "Techniques for Web
Content Accessibility Guidelines" [7] so that alternative presentation
of scripts can be available.

> 3.5. Image Maps
> Note for User Agents:
> It will be desirable that user agents render images and provide selections 
> with pointing (using area information) if possible. Even if rendering 
> images or selection with pointing are impossible, they should provide 
> access to linked resource of each area via "alt" attribute of AREA element.
> 
> HB: Image rendering is only possible in some mobile applications. Certainly
> not in the eyes-free environment. User should have choice. Why no mention of
> alt on img?

MI: This section only addresses the issue of "image maps", not the general
issue on images. That should be addressed in section 3.7.

> 3.6. Tables
> Note for User Agents:
> Even if it would be difficult, user agents may parse and render tables if
> possible. For example, the CAPTION element and "summary" attribute are 
> usable to summarize tables. See also [WAI-USERAGENT].
> 
> HB: Listifying a table is an alternative. So would user personality preference
> on how much expansion of TH information.

MI: Yes, and we think the general guidelines for table rendering should be
better addressed in User Agent Accessibility Guidelines, not the mobile
guidelines.  At the time of writing the mobile guidelines, we were
referring to WD-WAI-USERAGENT-19981112, which didn't have much
guidelines for table rendering, but now that WAI-USERAGENT-19990331
has more detailed guidelines [8], so I think we can just refer to
the relevant section of UA Guidelines in the next version.

> 3.7. Objects and Images
> Note for User Agents:
> User agents should try to render nested objects according to the precedence
> rules described in section 13.3.1 of [HTML40]. Even if user agents don't
> support OBJECT elements, they should render alternative content of OBJECT
> element as a fallback.
> 
> If a long description is provided via longdesc attribute of IMG element, user
> agents should provide access to that long description.
> 
> HB: Does that presume that otherwise it would handle alt="..."?

MI: Yes, maybe we should have mentioned about it for clarification.  In
section 3.7 [9], we state:

 "... alternative description via "alt" attribute, which is required
  in [HTML40], should be provided for user agents which cannot or do
  not render images."

It assumes that the user agents will handle alt attribute.

> 3.8. Access keys
> Note for User Agents:
> User agents should parse and activate access keys specified via "accesskey"
> attribute whenever possible, but they should ignore access keys when those 
> keys are not available.
> 
> HB: For some simple devices, only a few keys exist. Where does the user learn
> of the association of those keys (or keys in combination) with accesskey?

MI: We hope UA Guidelines will address this issue.  Just FYI, some
Web-enabled mobile phones only allow `0-9', `#' and `*' for access
keys, and they indicate the existence of these access keys with
specialized images.

Thanks again for your useful comments.


References

[1] http://www.w3.org/TR/REC-CSS2/generate.html#content
[2] http://www.w3.org/TR/1999/NOTE-html40-mobile-19990315/#Hardware
[3] http://www.w3.org/TR/REC-html32#script
[4] http://www.w3.org/TR/1999/NOTE-html40-mobile-19990315/#Styles
[5] http://www.w3.org/TR/NOTE-CCPP
[6] http://www.w3.org/TR/xhtml-modularization
[7] http://www.w3.org/TR/1999/WAI-WEBCONTENT-19990324/wai-pageauth-tech.html#scripts
[8] http://www.w3.org/TR/1999/WAI-USERAGENT-19990331/#gl-table-access
[9] http://www.w3.org/TR/1999/NOTE-html40-mobile-19990315/#Objects


Regards,
-- 
Masayasu Ishikawa / mimasa@w3.org
W3C - World Wide Web Consortium

Received on Saturday, 17 April 1999 06:00:15 UTC