Re: Path to Last Call (was closing various issues)

Hi Sam,

I have updates for you on some of the issues mentioned in this email  
which I hope you will find useful.

On Aug 22, 2009, at 5:17 PM, Sam Ruby wrote:

> Given that you said that issue 53 would only going to take a few  
> hours, if it needed to be addressed at all, I don't believe that it  
> is on the critical path, so I'm inclined to give Maciej and Julian a  
> bit more time to work on it.

Julian, Ian and I all seem to agree that ISSUE-53 needs some form of  
WG decision to resolve. I have proposed a lazy consensus resolution  
that I hope can resolve the issue. I'm not sure Julian and I need or  
want time to work on it - what we need is a WG decision.

> Issue 35 is still open.  I believe that for us to reach closure on  
> this will require someone like yourself or Maciej or Henri to  
> actually talk to someone like RichardS, StevenF, or Mike Cooper.   
> Yes, I understand that in theory this should be something that could  
> be accomplished via email, but in practice it simply has not happened.

Previously you said the next steps for ARIA integration[1] were:

A) The ARIA spec needs to enable host languages to make these  
distinctions [involving host language semantics].
B) Somebody needs to work out a matrix of potential combinations, for  
PFWG/WAI review.

Partial progress has been made on item (A). The PFWG posted their  
thoughts[2], but not yet full spec text, and indicated their  
willingness to collaborate on working out the details.

Item (B) has been completed. In fact, Ian went beyond the minimum  
requirements here and included the matrix and details of the  
conformance requirements in the spec, so the full consequences of his  
proposal are clear[3]. He also gave a list of questions regarding  
details he was unable to resolve based on the PFWG's input so far[4].  
This draft may not be perfect in every respect, but it may be  
sufficient to get us to Last Call.

I believe the key next step for ISSUE-35 and related is for PFWG to  
review the draft text and the set of issues Ian raised, to determine  
if there are any problems that they think block Last Call. Given the  
detail-oriented nature of any likely issues, I expect email is the  
best medium. However, if the PFWG would like to have some phone  
discussion, I am willing to facilitate and participate as needed. If  
no one raises objections to the draft text in the next two weeks, then  
I will consider ARIA-related issues resolved, and will propose that  
they be closed.

I believe another next step is for the PFWG to produce ARIA draft text  
incorporating their thinking on host language semantics, so we can  
verify that ARIA and HTML5 are consistent.

May I also add, it is not clear to me why you have continued to  
criticize Ian on ARIA. It seems to me that he did exactly what was  
asked, namely incorporate ARIA and engage with PFWG on the details.  
And it seems like further discussion is progressing nicely. Is there  
anything else you think he needs to do?

> I previously asked you if you could make the call of 13 August.  For  
> whatever reason, you did not attend.  I am now asking you if you  
> could work with Michael Cooper and/or Richard Schwerdtfeger to find  
> a mutual time in which you could participate, via phone, in a  
> discussion on what remains to be done to complete issue 74.  And,  
> yes, I am intentionally saying issue 74 as that is the one that at  
> this point in time looks like it has the most remaining work needed  
> to be done.  Issue 35 can proceed in parallel, perhaps even  
> exclusively over email.

This Friday, there was a conference call to discuss a proposal for  
canvas accessibility. I attended, as did Dave Singer and James Craig  
from Apple. I believe you were invited but for whatever reason decided  
not to attend. I know it can be hard to gauge the outcome of a telecon  
from minutes alone at times, so let me give you my own subjective view.

I believe there was rough agreement (if not yet full consensus) around  
an approach to canvas accessibility that, if it is found sufficient  
and effective, should be quite straightforward to integrate into the  
HTML5 draft. The basic approach being considered is to expose child  
elements of the canvas (and their possible ARIA roles and properties)  
to assistive technology, even though they are not rendered onscreen.  
Some select additional APIs may be needed for certain details, such as  
text caret/selection position.

The current next step is to prototype accessible versions of some  
Canvas content, to determine if the approach is workable and sound,  
and what further additions may be needed. We have volunteers to do  
that work. Based on prototyping experience, the group working on this  
will make a proposal.

I think more input from Ian would be welcome, but is not on the  
critical path at this time. This assessment is based on my personal  
involvement with work on this issue. I have also briefed Ian privately  
on progress so far. And I have asked the informal group working on  
this if they are willing to pull in the deadline for providing a  
proposal to the HTML WG, to some point earlier than December.

If you'd like to know any more about the status of this issue, feel  
free to ask.

> As can the closing of the other issues that Maciej is doing a  
> yeoman's job of clearing out the underbrush.  Ones that look, at  
> least to me, like they will require significantly less time than  
> resolving how to make canvas accessible.

I'm trying to apply the methodology of software project management. It  
is, of course, critical to monitor the critical path task (the  
"longest pole out of the tent") and do what is needed to ensure  
progress. At the same time, it's necessary to make progress on the  
smaller tasks. A project can just as easily be sunk by a flood of tiny  
issues, as by one huge issue. In fact, the former seems to be more  
common in practice.

In addition, I've discovered that the minimum time to close an HTML WG  
issue -- even one where there is no disagreement and no action is  
needed -- is about 10 days, including an average of 3 hours of effort  
by me (reviewing the issue, drafting email, following up on responses,  
attending at least one telecon, doing the tracker mechanics, etc.)  
Thus, even the most trivial of issues is going to take some time.

Conclusion: we can't put off the easy issues indefinitely. And I would  
appreciate support from the Chairs in getting them closed out.



Received on Sunday, 23 August 2009 04:06:29 UTC