Summary of the ERT WG face-to-face, October 2001
We met with the Protocols and Formats Working Group (PFWG). Part of our
discussion was Tuesday, the rest on Wednesday.
Present
(in person) Al Gilman, Charles McCathieNevile, Daniel Dardailler, Dave
Pawson, Wendy Chisholm, Libby Miller, Sean B Palmer
(on the phone) William Loughborough
Discussion
WC has been discussing the future of the ERT WG with a variety of people,
including the WAI domain team members. The consensus seems to be to move
forward according to the following plan:
- ERT WG takes 3-6 months to finish developing support materials for and
implementations of EARL. The primary future consumer and promoter of EARL
after the 3-6 months would be the QA activity. PF might have a hand in
this as well.
- The interface suggestions of the AERT are already incorporated into the
ATAG-TECHS. Future development of heuristics for identifying inaccessible
content and then repairing it should occur in the AUWG and documented in
ATAG-TECHS.
- The content "rules" of the AERT are being incorporated in the WCAG 2.0
Techniques for HTML as technology-specific checkpoints. Future development
of rules should occur in the WCAG WG and be documented in
technology-specific checkpoints.
- The list of evaluation and repair tools would be maintained by the AUWG
as part of their list of authoring tools.
The concerns with this proposal are:
- that a specific chapter or document focused on evaluation and repair
exist for those who are only interested in evaluation and repair (such as
CAST, ATRC, others).
- there should still exist an area for people to chat to discuss ERT
Issues
Suggestions for addressing these concerns
- Since ATAG Guideline 4 is
solely about checking and correcting inaccessible content, it was felt
that it would be easy to extract this info for the ERT-specific
audience.
- It was proposed that the ERT morph from a working group into an interest
group.
Those present then felt that it was appropriate to close the working group.
The issue of morphing from a working group to an interest group will be
discussed further. The primary focus of the upcoming months is to create
documentation for EARL as well as implementations (proof of concepts). We will
also continue to brainstorm about possible producers and consumers of EARL and
how we might "sell" it to them.
Present
(in person) Al Gilman, Charles McCathieNevile, Dave Pawson, Wendy Chisholm,
Libby Miller, Sean B Palmer, Nick Kew
(on the phone) William Loughborough
Discussion
We brainstormed about a variety of use cases for EARL. Who are possible
producers and consumers? What types of materials do we need to develop for
them?
- Comparison of results from tools (developer)
- Looking for specific content that meets specific WCAG checkpoints
(reader/consumer of Web material)
- Tech support for solving interoperability problems between tools. For
example, a users configuration that requires an assistive technology. When
trying to figure out where a problem occurs in the system, have to look
at: assistive technology, operating system, other applications, etc. Users
need a data package, that details version, settings, stream of commands.
In other words, capture as much info as possible to send to tech support
to give them the specifics of the problem. Consumer - help desk. Need -
tech support interoperability support. Producer - user.
- 2nd opinion brokering
- site-evaluation groups, comparing results between members of the
groups.
- across disability groups - input into WCAG
Present
(in person) Al Gilman, Charles McCathieNevile, Dave Pawson, Wendy Chisholm,
Libby Miller, Sean B Palmer, Nick Kew
(on the phone) William Loughborough
Discussion
Dave Pawson made the case that people need to see a complete end-to-end
process and how EARL fits in. He drew a diagram with the following steps:
- Requirements capture
- Test specification
- Test
- Test result
- Collate results
- Analyze results (query)
- Present results
EARL is the method used to store the test results. It was also suggested
that the test specification should also be machine-readable and WC took an
action to investigate developments in this area.
The end-to-end process is possible (when using EARL) because the bits are
not proprietary. This is the beauty.
Tools should be developed to show (proof-of-concept) how this works. It
should also be described in prose. Both of these should help to sell the use
of EARL to potential producers and consumers.
Present
(in person) Al Gilman, Charles McCathieNevile, Dave Pawson, Wendy Chisholm,
Libby Miller, Sean B Palmer, Nick Kew
(on the phone) William Loughborough, Harvey Bingham, Katie Haritos-Shea
Discussion
We brainstormed about the various tools that could be created to produce or
cosume EARL and who was interested in working on or helping come into
being.
- EARL in authoring tools. e.g., ACCRepair, A-Prompt - more than one tool
sharing results of evaluation and repair info.
- Assessing usability of assistive technologies (AG)
- Authoring tool reviews - convert to EARL. (CMN)
- SiteValet produce EARL (Nick will work with Sean)
- W3C service-reporting
- Libby is willing to work on a simple query if we supply her with data
and the sorts of questions we might ask. (LM)
- EARL Python API - encourage semweb folks to get involved (SBP)
- Sales story to sembweb folk
- Accessibility review of W3C site. Systeam wants to automate process of
getting comments. SiteValet use to benchmark. DOG FOOD. Accumulate log of
issues and resolutions in EARL - data for Libby (AG)
- WCAG test suite cleared by WCAG. Then the tests will be used by ATAG for
evaluations. Use ATR to do so. Generate more data for Libby. The growing
WART. (Charles Munat, WC, CMN)
- W3C validators produce EARL. (CMN's hopeful INRIA student)
- WAVE-type evaluations. Human-based testing. Does alt-text make sense?
Person able to say "yes" and not be asked again. A-prompt and AccRepair
etc could also do (AG)
- Schematron produce EARL
- Requirements capture. Report and analysis. XML to presentation. i.e.
take EARL and make it into something humanly useful. (DP)
- RNIB survey of UK web sites. convert data to EARL to analyze? (DP and NK
took an action to investigate)
- WCAG - UI Key. HPR hit F9 (or some key) that pops up a form. A plug-in
for IE6 or 7. Enables user to report problems without effort. The data -
catalog of accessible content.
- Put EARL conformance claims to WCAG in a database. When someone does a
Google search you can cross-reference between the Google database and the
WCAG conformance database to find content that meets your needs.
Particularly useful for people with learning and reading disabilities.
(WC)
- XAG Evaluation in EARL to prepare CR implementation report. (CMN -
extension of WART)
- QA focusing on W3C-owned stuff - dog food patrol. We should provide them
with tools. (WL)
- WAVE report in SVG. SVG give metadata place and visual displays of bug
flags. Query - get context menu. EARL included in metadata in SVG. Test
case for John Farrialo's use of forms in SVG. Richer than current WAVE
interface. XForms, SVG, EARL, WAVE. (AG)
- Annotea - interface to generate EARL as annotation instead of HTML
(CMN)
- EARL output on 10 people's evaluation of alt-text differ. This info can
tell you how fluffy the alt-text test description is. Good data for WCAG
as trying to develop testable checkpoints. (DP)
- DAISY - collation of outcoe of AT for DAISY OK stamp. Schematron the
alternative. Earlier mentioned talking with Rick about Schematron
producing EARL. (DP and Harvey Bingham)
- How to sell the idea of publishing EARL data about their tools to folks
who don't conform? Person using has to make things that work well.
- Publicize ATR results? talk with developers. Make it into a consumer
reports of ERTs? Limiting scope of damage knowing ATAG quirks.
- Testpoint definition language make the end-to-end possible. Companion to
EARL. RDFSchema, SemWeb, logic processing. (DP) What is NIST doing?
(WC)
- We will work to make these tool ideas come to life.
- Figure out how we will fit into QA, AUWG, WCAG WG, and PFWG.
- Determine if we will continue as an interest group.
- Illustrate an end-to-end process in prose and also hopefully with a set
of tools and data.
- Begin to transition the WG to either an IG or non-existent in the next
3-6 months.
$Date: 2001/10/08 21:50:24 $ Wendy Chisholm