Re: feedback requested on WAI CG Consensus Resolutions on Text alternatives in HTML 5 document

Hi Ian,

On Aug 19, 2009, at 1:12 PM, Ian Hickson wrote:

> On Wed, 19 Aug 2009, Sam Ruby wrote:
>> Ian Hickson wrote:
>> I understand that Steven has proposed a set of rules for validators  
>> to
>> use which he believes will encourage the adoption of alt  
>> attributes.  I
>> further understand that Ian has draft text which proposes a different
>> set of rules for validators to use which he believes will encourage  
>> the
>> adoption of alt attributes.  I understand both sets of rules.
>> As to whether or not I understand the rationale for each: frankly no.
>> Actually, that's not quite right.  I've read the words in each case,
>> believe that you both are being a bit, shall we say, optimistic.  But
>> that's just me.
> I think understanding the problem that Steven's proposal solves is an
> important part of evaluating Steven's proposal, and a critical part  
> of me
> merging it into HTML5.
> I find it ridiculous that you suggest that my admitting not  
> understanding
> the proposal's rationale is somehow a bad thing when you yourself now
> admit to not understanding the proposal's rationale.

I think Sam was absolutely long to criticize you and others who asked  
for rationale. First, many aspects of the proposal were not at all  
clear without digging deeper. The goals of many of the suggestions  
were certainly not clear to me, and apparently not clear to Sam  
either. Second, we've actually had some responses which explain the  
background of the proposal in a way that sheds light on the issues. So  
I think asking for rationale, as you did, was absolutely the right  
thing to do.

> Steven, if you could describe for me the problem that exists in the  
> spec that your proposal solves, I would be more than happy to  
> address said
> problem, and would be grateful for your proposal.
> Without a description of a problem, however, I do not intend to edit  
> the
> spec on this topic.

I believe Steven has now given much of the needed explanation.

A) The Consensus Resolutions document suggests that alt="" (empty alt)  
without role="presentation" on the same element should trigger a non- 
fatal validator warning that recommends adding role="presentation".
Steve explained here and in follow-up emails, some of which questioned  
the given rationale and suggested the warning might not be a good idea  
after all:

B) ARIA markup should let you completely omit "alt":

After examining the reasoning, Steve agreed that this should not be  
allowed, at least in the case of images as the sole content of a link.

C) title / sole-image-in-paragraph exceptions not allowed as  
exceptions in the case of unknown image contents:

Steve explained here that this is because title does not render like  
alt with images disabled or in text-only browsers:

Henri added that autogenerated title would possibly violate the spirit  
of ATAG2:

Jan Richards suggested a "missing" marker as a way to flag images with  
deliberately omitted alt, so that autogenerated descriptive text would  
not be necessary but conformance checkers could continue to flag  
errors. Henri agreed this might be a viable way to resolve the seeming  
conflict between HTML5 and ATAG2 requirements:

D) Omission of the "private communications" exception and detailed  
examples was not a request to remove those.

E) Requested reference to WCAG:

Steve gave some explanation here and drew comparisons to HTML4 and SVG:

I don't necessarily agree with these rationales, but perhaps they will  
be sufficient for you to comment on the issues. I think Steve and  
others providing these shed much-needed light on the underlying goals.


Received on Thursday, 20 August 2009 02:36:09 UTC