W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2001

RE: Bulk use of accessibility checkers and other auditing tool

From: Al Gilman <asgilman@iamdigex.net>
Date: Fri, 03 Aug 2001 12:08:53 -0400
Message-Id: <200108031553.LAA4211318@smtp2.mail.iamworld.net>
To: Brian Kelly <b.kelly@ukoln.ac.uk>, "'Charles McCathieNevile'" <charles@w3.org>
Cc: w3c-wai-ig@w3.org, chitchcock@cast.org
At 11:47 AM 2001-08-03 , Brian Kelly wrote:
>Thanks Charles.  I was aware of the EARL work, and it has some
>relevance. However although it will allow you to make statements such as
>"The entry point is 50K which is within the x guidelines" in a
>machinereadable way, the meaning of the size of a page does not yet
>appear to be defined anywhere - i.e. does it mean the orgin of a
>redirect page, the destination, or the sum; what does it mean if
>user-agent content negotiation occurs; etc.

AG::

Brian,

This is an excellent issue.  Just please put an arm's length between "the size
of the entry transaction" and the phrase "the size of a page."

What we need to define is "that to which the data volume standard applies." 
This is the size of the "enter the site transaction."  Not any single
management unit of retained data.  The page notion is handy in the user
appreciation of what they do and get, but generates confusion when we think we
can bind it one-to-one onto something the service retains or does.

Talking about "the size of a page" puts you unfortunately in static "format"
view whereas for measuring network load you want to be in protocol or
"transaction" view.

The unit you want in the rule you mention is 

  The total data communication which is automatically engendered when the user
activates 
  a hyperlink referencing "the home page" of "a site." 

That's almost right.  First variation is to exclude time-delayed
auto-refreshes.  Second variation is to include back in time-delayed
auto-refreshes of under [say, ten seconds] delay so that automatically passing
through a splash page is counted as an automatic part of what you have to go
through on entering the site.

But these details should be handed as definitions first and standards second
(later).

What constitutes "a site" is up to the service offeror.  We presume that there
is a default starting URI for a given site and that is "the home page" for
that
site.

The best commercial practice here is that this is something with an http: URL
that is all DNS and no filepath.  But our assessment vocuabulary should allow
for the units of service, and home URI spelling, of the offeror's choice.

Your rule will constrain the volume of network data transport service that the
user has bought by one "explicit user action" event, in the idea-set of the
UAAG.

All the variations in the actual network utilization episode that you mention
show that the actual data volume is a distribution of different values under
various contingencies.  But if there is an automatic redirect, the traffic for
both GETs is clearly included.  For negotiated alternatives, you have to pick
your policy.  There is no policy-free technical definition.  In terms of
surveys, standards-setting, and quality of service measurement, it is
reasonable to use averages across the experienced distribution of hit-subcases
as "what is constrained by the norm."  Perhaps max limits as well.  What you
set norms for is there a policy matter, not a matter of common definitions
that
can serve a variety of policy decisions.  The technical terms are the
primitives that you use in defining what the norm applies to, not the bottom
line you decide to constrain.

>I'm not sure who should be involved in this work.  Is it within scope of
>the WAI group?  If not, is there any work going on in, say, the Web
>auditing communities - I know there is some standarisation work going on
>in that community.
>

Any standardisation work in Web auditing would be of interest to at least EO
and ER.  If you can provide pointers where we can follow up that would be
super.

This topic intersects with the scopes of at least ER, GL, and EO.  GL is
endeavoring to test the testability of their pronouncements before publishing
them as guidance.  "Don't make your home page too big" is a web content
guideline, IMHO.

Al

>Thanks
>
>Brian
>---------------------------------------
>Brian Kelly
>UK Web Focus
>UKOLN
>University of Bath 
>BATH
>BA2 7AY
>Email: B.Kelly@ukoln.ac.uk
>Web: <http://www.ukoln.ac.uk/>http://www.ukoln.ac.uk/
>Phone: 01225 323943
>
>> -----Original Message-----
>> From: Charles McCathieNevile
[<mailto:charles@w3.org%5D>mailto:charles@w3.org] 
>> Sent: 03 August 2001 14:03
>> To: Brian Kelly
>> Cc: w3c-wai-ig@w3.org; chitchcock@cast.org
>> Subject: Re: Bulk use of accessibility checkers and other 
>> auditing tool
>> 
>> 
>> Hi Brian,
>> 
>> there are checkers that are starting to offer customisation 
>> features. The ones I know are either ones that cst money, or 
>> require other tools that cost money.
>> 
>> But finally, yes, there are tools producing output in a 
>> machine readable format, there is work in the ERT group on a 
>> project called EARL (Evaluation and Report Language) which is 
>> an RDF vocabulary specially designed for reporting results 
>> like Bobby in machine readable format. And Josh Krieger and 
>> CHris Ridpath (noted for Bobby and A-prompt respectively) 
>> have just  produced a tool that outputs EARL.
>> 
>> More information is probably most easily available from the 
>> EARL homepage
<http://www.w3.org/2001/03/earl>http://www.w3.org/2001/03/earl
for people 
>> intersted in the details, or from the ER group's mailing list 
>> archives - look for example at the thread beginning at 
>>
<http://lists.w3.org/Archives/Public/w3c-wai->http://lists.w3.org/Archives/P
ublic/w3c-wai-> er-ig/2001Jul/0055
>> 
>> cheers
>> 
>> Charles
>> 
>> On Fri, 3 Aug 2001, Brian Kelly wrote:
>> 
>>   Has anyone provided standard definitions for what 
>> constitutes a page,
>>   and what action user-agent should take when such strange 
>> things happen?
>>   Are any auditing tools providing customisation over the actions they
>>   will take?
>> 
>>   Finally are any of the Web sites which provide such Web analysis
>>   features looking at going down the "Web service" route, and 
>> providing
>>   output in a machine understandable format - so that the 
>> results can be
>>   more easily post-processed?
>> 
>> 
>> 
>  
Received on Friday, 3 August 2001 11:57:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:13:55 GMT