Re: LC#127: How to verify 5.7 (Provide programmatic exchange of information in a timely manner.)?

hi Jon:

At 4:47 PM 1/19/00, Jon Gunderson wrote:
>I propose that we discuss this in the techniques document related to this
>checkpoint.  Probably the easiest thing to do is to state programming
>examples on major plateforms that would be acceptable.
>
>Some technique issues:
>1. In process versus out of process access
>2. Microsoft COM technologies
>3. COBRA technologies
>4. Mac???
>5. UNIX???
>
>I think we will need some programming specialists to help fill out this
>list.  Maybe Rich Schwerdtfeger and Mark Novak could help out and maybe
>talking to some of the people in the Amaya team.


I noted this was still a "continued" action item from yesterday's UA mtg
(thursday), so I thought I'd try to toss out some ideas on this issue.

Let me start by saying I fully agree that we must have timely access
to the exchange of information.  What I stumble on, is how to "verify"
the timely wording.

First, without checking the list archives, I know this issue of "timely
access" has been around for a long time.  As I recall, this was a question
which repeatedly was not satisfied when people were doing evaluaitons of
the early guidelines against current browsers.  Interestingly enough,
when I asked those reviewers what criteria they used for making this
statement, I heard everything from "no response"  to "well, so and so
said it was".  In other words, nothing quantitative.  This doesn't
surprise me.  I would expect that we each have a different concept
of what "timely" access would be.  Some of us may describe this
as "immediate" while others may have a perfectly usable UA/AT
that allowed a few seconds to lapse.  I don't think anyone knows what the
correct "value" might be that constitutes "timely", thus I'm concerned of how
we will "verify" this.

I know the system responds much faster (information exchange), when switching
from an out-of process application which used COM to access the
IE DOM as compared to an in-process application which used COM (e.g.,
also referred to as a browser helper object [BHO]).  I also recall Rich
sent to the list some similar results they had quantified while testing
at IBM.   The technology used to access the information which needs to
be exchanged, can be done in any number of methods.  Some of these methods
you have listed above, of which these and others are already in the
Techniques DOC, in Appendix 5.  AT developers will have to choose
which "method" best matches the needs of their technology and users, and
yes, there will be trade-offs in timeliness, robustness, reliability,
etc., for which again, I'd expect the AT and perhaps the UA developers
to understand and work with.  I think what is important in all this is
that there "is" a way to get the information exchanged, and provided
a AT or UA developer use the most current technology available for
the platform(s) they wish to support, then they have probably given
the user "timely" access.  If not, then I suspect the product won't
survive, since the users won't be satisfied and thus will or should go
somewhere else for a solution (I realize this isn't always possible).

Can we do better than this to"verify" timely access.  Perhaps.  The
one other thought I had was to provide a "reference" design, maybe
as part of the Techniques DOC., or better yet as a tool off the ER page.
What I'm envisioning the "reference" design would be, is a fully
functional prototype that when run on a "very specific" platform
and operating system, would demonstrate the "timely" response to
making whatever information we deem necessary which shows how a
UA/AT ought to respond.  Then at least the UA group could say to developers,
this "reference" design demonstrates what we feel is necessary.

Before everyone decides this is a good idea, realize this will take a
considerable amount of work, and we would still have to "quantify" timely
access, and what information we'd exchange/show and many other
details, messy details.  ;)

Anyway, just my quick thoughts.

Regards

Mark

PS:  Note, there already exits in the Techniques DOC., a URL for
a BHO example, under Appendix 5 also.

Received on Friday, 21 January 2000 14:27:47 UTC