Re: Proposed rewording of the bundling clause (draft)

Hi Tim,

Your detailed feedback is much appreciated and has led to some important 
changes – especially with regard to what requirements needs to be done by what 
tools in the process. I will try to address your comments in-line. (I have also 
tried to edit all this for succinctness)

Quoting Tim Boland <frederick.boland@nist.gov>:

> I have concerns with the proposal, in relation to the following QA SpecGL
> [1] normative requirements:
> 
> QA SpecGL Requirement: define the scope - I think that the proposal may 
> have scope implications, and so may need to be addressed in the ATAG "scope" 
section

JR: Agreed.

> QA SpecGL Requirement: Define the terms used in the normative parts of the
> specification - I think that ATAG may need some additional definitions and 
> consistent usage as indicated following

JR: Agreed.

> QA SpecGL Requirement: Create conformance labels for each part of the 
> conformance model - as indicated following, I have difficulty understanding 
the 
> conformance model being proposed (perhaps a diagram would help?), and how 
> such a conformance model so constituted would relate to ATAG A, AA, or AAA..
> The conformance model being proposed seems to me somewhat complicated and
> confusing.. I think that an unambiguous objective definition/criteria/rules 
of 
> bundled vs. nonbundled might help?

JR: I may not have explained the idea properly. There is one conformance model. 
It applies to the “process” by which tools help authors to create, edit, and 
publish Web content. (of course the create, edit and publish steps will have 
many permutations, depending on formats involved, sophistication of  content, 
degree of automation, etc.). A particular tool (A) may perform the entire 
process, in which case an evaluator can undertake an ATAG evaluation just on 
itself. On the other hand, a tool (B) may perform only part of the process, in 
which case an evaluator can identify the other tools (C and D) that make up the 
process and undertake an ATAG evaluation on all of the tools together (B and C 
and D). Now to avoid dependence on particular tools that do other things in the 
process, it is possible to identify two or more equivalent tools that can be 
substituted for each other (B and C and (D or E or F)).

> QA SpecGL Requirement: If the technology is subdivided, then indicate which
> subdivisions are mandatory for conformance - I think that there may be some
> implications on this requirement in proposal..

JR: I don’t think so, because all parts are mandatory for the process as a 
whole. The accessible authoring interface requirement must be met by all of the 
tools in the process, but this also does not seem to me to be a subdivision. 
(There has been some interest in developing a proposal for developing different 
requirements for the tools and the process – but that is not this proposal).

> QA SpecGL Requirement: If the technology is subdivided, then address 
> subdivision constraints- I think that there may be some implications on this 
> requirement in proposal

JR: Once again, I don’t think this applies.

> More specific in-line comments following:
> My apologies if I have misunderstood the proposal.
> Thanks and best wishes
> Tim Boland NIST
> 
> [1]: http://www.w3.org/TR/2004/WD-qaframe-spec-20041122/
> 
> At 10:32 AM 3/9/2005 -0500, you wrote:
>
> >Here is a draft proposal for reworking the bundling clause as a a
> "process"
> >clause. (item (1) from Monday's call) (this proposal would not effect any
> >priorities, etc. - that is a different work item)
> >Thoughts? Developer perspective? QA perspective?
> >Cheers,
> >Jan
> >
> >PROPOSED NEW WORDING:
> >
> >3.2.2 Authoring Tools as part of an ATAG Conformant Process
> 
> What is an "ATAG-conformant process" (needs to be defined)?  What is a
> "process workflow" (needs to be defined-workflow is defined in
> the glossary alone, but "process" is not)?

JR: I intended for that term to be made clear in the text that followed. 
Thinking about this further, I think that if we take a “process” approach that 
has to be made much more clear throughout the doc.

> >The requirements of ATAG cover all aspects of the process by which
> authoring
> 
> What is a "process"?

JR: The definition of “process” is something like: the sequence of tool-
assisted steps that authors use to create, edit, and publish Web content. (e.g. 
a simple process: enter text (text editor), edit images (image editor), create 
style sheet (style sheet editor), markup structure (markup tool), insert images 
(markup tool), apply style sheet (markup tool), check for and repair 
accessibility problems (accessibility evaluation and repair tool), and validate 
syntax (validation utility). 

> >tools aid authors in the production of Web content. In some cases, the
> >entire process will take place within a single comprehensive authoring tool, 
> >while in other cases, the process will involve two or more authoring tools 
used 
> >together
> 
> What is the definition of a "single comprehensive authoring tool" as opposed
> to "separate authoring tools"?  Are these two different "classes of products"
> that would implement ATAG differently, and have different ATAG conformance 
> requirements? It seems that in one case, a (process?) is "contained" within 
an 
> authoring tool, whereas in the other case,  a process "contains"
> multiple authoring tools.    To me, if this is correct, the relationship of
> process to authoring tool seems somewhat convoluted.. I'm having difficulty 
> understanding the ATAG conformance model being suggested..

JR: Yes that is an inconsistency. What I meant is that the process, as defined 
above, is a sequence of steps. One or more authoring tools will assist the 
author with the steps.

> >(e.g. a markup editor with plug-ins) or separately (e.g. a markup editor,
> >an image editor, a validation tool, etc.). In either case, any of the
> >authoring tools may claim ATAG conformance provided that a description 
> >of the process workflow is provided (link: future example), and that the 
> >process workflow meets and documents the following conditions:
> 
> Must the description of the process workflow provided to the author in 
> advance of use of said workflow?

JR: The process workflow description is only provided with the conformance 
claim. It informs the purchaser that the tool is part of an ATAG conformant 
process (either by itself or with an identified list of other tools).

> What are you testing here, the process workflow or the authoring tool?  Say
> "(all of) the following conditions.."?

JR: We are gathering together all of the tools listed in the process 
description and testing whether (together) they meet the conditions of ATAG.

> What is a "process
> workflow" (needs to be defined-workflow is defined in the glossary alone, 
> but "process" is not)?

JR: I suppose we could drop “workflow” and just use the definition of “process” 
that I provide above. 

> >(a) All authoring tools must be identified by name. If more than one 
> >authoring tool can be used at a particular point in the process (e.g. for 
> >image editing or accessibility evaluation and repair, etc.), then the
> >description may identify two or more interchangeable tools.
> 
> "All authoring tools (used in the process workflow) must be.."?   Use 
> "must" instead of "may"
> (is the second sentence a normative requirement (objectively testable)?

JR: I think it is “may”. For example, let’s say an evaluator is in the 
marketing department of the developer of stylesheet editor. In order to make an 
ATAG claim for the company’s stylesheet editor, the evaluator must show that 
the editor can be part of an ATAG conformant process. To this end, the 
evaluator documents a process that involves the company’s stylesheet editor as 
well as a particular markup tool and an accessibility evaluation and repair 
tool. Testing is done, and it turns out that the combination of tools meets 
ATAG level A. That is all that is required. Even though the evaluator has the 
option of identifying other markup and evaluation and repair tools that could 
be swapped in and the same ATAG conformance level achieved, there is no 
requirement to do that.

> >(b) All of the authoring tools in the process workflow must meet the 
> >accessible authoring interface requirements (for the desired conformance 
level) 
> >(link: the checkpoints for guideline 1).
> >
> >(c) At least one of the authoring tools must meet each of the accessible
> >content production requirements (for the desired conformance level).
> 
> Does one specified authoring tool need to meet all of the accessible.., or
> can different authoring tools meet different requirements (for example, 
> authoring tool a
> meets requirement c, authoring tool b meets requirement d, etc.)?

JR: Yes, that is what I meant. But you have made me take a closer look, and it 
seems that some of the requirements covered by (c) actually should be required 
in all tools (and this leads towards that other proposal about a split between 
things required by all the tools in the process and things that just need to be 
in the process). A possible breakdown:

All tools: 1.1, 1.2, 1.3, 1.4, 1.5, 2.1, 2.2, 2.3, 2.4, 3.1, 3.4, 3.7, 4.1, 
4.3, 4.4
At least one tool in process: 3.2, 3.3, 3.5, 3.6, 3.8, 3.9, 4.2

> >(d) All tools that do not implement the checking (link: 3.2) and repair 
> >(link: 3.3) requirements must occur earlier in the process workflow than
> >at least one authoring tool that does implement these requirements.

> All (authoring) tools..  Also possible testability problems with "proving a
> negative". The reference of this requirement to other requirements may further
> complicate the conformance model..

JR: Agreed. All I was trying to say is that it is no use having checking and 
repair come before more content creation.


> >OLD WORDING:
> >
> >3.2.2 Bundling Authoring Tools
> >A conformance claim may cover more than one authoring tool used together 
> >(e.g.
> >a markup editor and an evaluation and repair tool or a multimedia editor 
> >with a
> >custom plug-in), each of which may or may not conform to ATAG 2.0 by
> >themselves, as long as:
> >
> >- The authoring tools are distributed together.
> >- The workflow used to determine the conformance of the tool bundle is 
> >typical
> >of how the tools are used together.

-- 
Jan Richards, User Interface Design Specialist 
Adaptive Technology Resource Centre (ATRC), University of Toronto 

  Email: jan.richards@utoronto.ca 
  Web:   http://jan.atrc.utoronto.ca
  Phone: 416-946-7060 
  Fax:   416-971-2896

Received on Sunday, 13 March 2005 02:54:29 UTC