[Accessibility] Meeting minutes 9/17/2003

Sharon D Snider snidersd at us.ibm.com
Sat Sep 20 15:15:27 PDT 2003





Meeting minutes 9/17/2003 draft

A meeting of the Accessibility Working Group was held
by teleconference on Wednesday, Sept. 17, 2003, at
18:00 UTC

Attending:
Doug Beattie, Free Standards Board
Randy Horwitz, IBM
Janina Sajka, American Foundation for the Blind (AFB)
Gunnar Schmidt, KDE
Sharon Snider, IBM
Matthew Wilcox, HP, Linux kernel developer group
Alan Wilson, IBM
Bill Haneman, Sun

Action Items:
John - Check to see if Access utilities are patented by U. Wisconsin.
Doug - LSB test information and Web sites for TET.

Janina updated everyone on the status of FSG102:
- FSG102 has been submitted to the Board of Directors (BOD).
- Scott requested grammatical changes prior to posting it to the board.
- Doug requested Section 3 be added prior to submission.
- Voice I/O not itemized.
- Next Monday, 9/22/2003 BOD will vote on FSG102 proposal.

Next Steps:
- FSG expectations for test and certification process.
- Who can we recruit for Subcommittee?

Doug - LSB uses the TET framework that verifies an application, and
assembles a document that reports failures, warnings, and successes. The
report is validated to confirm it has not been tampered with. Chris at LSB
works with test suites that can possibly be used for accessibility testing.

Janina - We can test that the ABIs don't break with automated tests, but
human intervention is required for query fields and functionality. It takes
a human to verify accessibility. For example, "ALT TEXT", what happens when
a button is pressed? A visual description is not needed, but the action
information is.

Bill - Focus should be on ABI, API, and functional. Light on the
functional. We need to verify that the platform provides an infrastructure,
but not that it has been used correctly. The test should confirm the
platform provides all the APIs and binary signatures. Tests that verify
APIs exist and work correctly is a big job in front of us.

Doug - We can consider validating distributions to confirm that the OS
provides required tools, and validate that the application is not using
anything outside what the LSB says should be provided. There is an LSB app
checker that identifies calls outside of the spec. Its acceptable as long
as it is provided in their own name space. I'll send pointers on what we
need to be doing for binary compatibility.

Janina - We should be testing at the same level as other LSB tests and
certifications.

Bill - Provide certification with some phases not being available yet.

Janina - What is the process? Who can be part of the Subcommittee? White
papers will be helpful and how to apply the tools.

Bill - Verification of ABIs might be in line with what he's doing. He
should be involved in the subcommittee because of his familiarity with the
ABIs and APIs.

Doug - To help facilitate LSB testing, they have been looking for ways to
find partners and Universities to run validation and test suites. Templates
would be provided to the students for implementation. There are multiple
ways to approach testing.

Janina - Will discuss with Neil and Greg.

Bill - Its hard to build a research project around ABI verification.
Corporate partners may have an interest.

Doug - Sun, HP, and IBM would probably like to know that an application is
certifiable and be able to write applications and know they will run
without difficulty.

Bill - 508 gives neither API or ABI testing. It is only functional. ABI and
API is less controversial.

Allen -  Needs a schedule and time before he knows when and how involved he
can get. Does anyone have experience in estimating the resources needed?
How useful is it to only test interfaces without functionality? We should
take a phased approach. The keyboard can't be tested with an ABI.

Bill - How much of the effort can be off loaded to applicants?

Janina - Should there be someone outside the company to run the test?

Doug - No, the applicant that runs the test signs their name to the
results. Part of the process is if there is a problem, the applicant will
fix it. Applicants certify, in writing specific information. If they are
not compliant an applicant can lose certification.

Randy - Is this comparable to J2EE certification?

Bill - There are some areas we should certify that do not lend itself to
automation. We could supply the applicant with instructions and an
automated test suite. A checklist, write test ascertains up and the
applicant can manually complete them. Verification is only as good as the
tests.

Allen - Are we discussing certification or verification? From a legal
perspective there is a difference.

Doug - a lot of this has been covered and dealt with by the FSG.

(See attached file: meeting_minutes_20030917.txt)

Sharon Snider
Linux Accessibility
IBM Linux Technology Services
(512) 838-4127, T/L 678-4127
-------------- next part --------------
A non-text attachment was scrubbed...
Name: meeting_minutes_20030917.txt
Type: application/octet-stream
Size: 4742 bytes
Desc: not available
Url : http://lists.linux-foundation.org/pipermail/accessibility/attachments/20030920/ea609b05/meeting_minutes_20030917.obj


More information about the Accessibility mailing list