1 May 2000 ER WG Telecon

Summary of action items and resolutions


Harvey Bingham, HB



WC: talking to people; as of Friday, have 14 people registered for Fri/ 13 for Thursday; includes some new people; found out UsableNet URI and tool -- both developers will be at meeting (from Italy); contacted everyone on list of existing tools; heard back from a few with conflicts; people commenting that project interesting; Rick Jelliffe could participate via phone, but not physically -- what do you think about that?

LK: would be great if can take care of logistics

WC: also said would send a position paper if that would help; will follow up; Authoring Tool People: CMN, Jan,

DB: I'm not coming, but Cynthia Shelly is

WC: Marjolein Katsma from Allaire, maybe someone from Adobe

LK: what about the Edapta people; met president Jack Berkowitz at a meeting last week -- interested in joining ER, going to ask him to join; have a tool that presents web site in different formats: most abstract info in XML, style into different forms; other extreme ordinary visually oriented HTML page, with additional stuff to accommodate browsers

// Action LK: ping Edapta about participation in ER //

HB: not attempting to adapt itself via browser sniffing?

LK: no; uses data profiles;

// DD joins momentarily to hold phone for WC //

GJR: what kind of data profiles? CC/PP or rolling their own

LK: not sure, Kynn Bartlett, formerly of HTML Writers' Guild (HWG) involved

GJR: with Kynn Bartlett there, probably at least investigating CC/PP, as that is one of his special interests

LK: Daniel, talking about whom we have invited outside group to participate in F2F

// WC rejoins, DD leaves //

WC: hope more people registered over weekend;

LK: any additional agenda item suggestions? last week took agenda from web site and made one major change--in the pre- lunch timeslot put in strategizing and planning, and added 15 minutes after lunch to debrief group with thoughts and ideas that came up during individual lunchtime discussions

WC: Charles McN suggested talk about charter; don't think necessary; think agenda y'all came up with last week ok; on Friday, a lot of discussion of how AU and ER will work together

LK: Thursday conversation involves charter issues even if not charter discussion per se


LK: has everyone access to charter? if not, can read as I go

[Scribe's Note: Proposed ER Charter can be found at: <http://www.w3.org/WAI/ER/erwg-charter.html#mission>]

// LK will read as he goes //

WC: Charles MCN had some issues I haven't had time to add to draft

LK: key aspects of this is the mission and deliverables, especially in terms of preparing for F2F--let me review the Mission Statement now: The mission of the Evaluation and Repair Tools Working Group (ER WG) is:

  1. to document techniques for creating Evaluation and Repair Tools;
  2. to find tools that implement the techniques and where there are none, prototype or participate in the development of an implementation;
  3. to assess the implementation of these techniques in evaluation and repair tools;
  4. to provide a discussion forum to review and collaborate on tool development.

// Brian Metheny (BM) joins //

GJR: are we doing away with the IG?

WC: yes; authoring tools WG and WCAG and UA all provide function of what was defined in interest group; need to focus on creating tools and causing them to come into being; part of that is forum for developers; encouraging people to implement ideas and share ideas and implementation strategies

GJR: don't' disagree with strategy, but think that the ER activity's proper purview is developing tools, not theorizing about them; consciousness raising is EO's purview

LK: if charter goes through, what sorts of messages would we see in archives one year from now?

WC: personally, hope to see more discussion of techniques, as when we went through techniques to push AERT out the door; want to see people talk about tools--I'm developing this tool, this is what I'm checking for, this is what I found, this is how I repair; someone else says, I do it this way, someone else says I took X and Y and then integrated it into tool Z; want to try to cross-fertilize commercial developers participating in AU WG with our ideas

LK: when people discuss techniques, how deep into implementation will they go? high level of abstraction, or greater? Perl allows you just to get the <HEAD> of an HTML document, so could use Perl script to grab the TITLE of a page to which a graphical hyperlink without ALT text points

WC: can have a variety; continue to find questions to send back to WCAG or UA WG

GJR: again, I don't disagree with strategy, but think that the ER activity's proper purview is developing tools, not theorizing about them, take strategies outlined by IG and other guidelines documents and implement them as proxies to find out (a) what works, (b) for whom, (c) under what circumstances, and then we can (1) take that empirical evidence to developers and ask for implementations, and, equally as important in my mind, keep the proxies available for those who don't have access to the latest and greatest software and hardware

LK: when we talk about "we", who are we talking about? in terms of talking about developing techniques; the way the group works now, is that there are several people who happen to be part of this group that develop these types of tools; don't have a group activity other than ERT or guidelines

GJR: like to see us go further down the path of Tablin, the report tool, etc.

WC: how do you see that playing out practically?

GJR: to make most difference in the real world for actual users, we need to test ideas, rather than formulate them; there may still be a need for an IG and a WG--IG formulates techniques, based on other WAI WG GLs, then WG implements them in a variety of ways, as proxies; then we can say to the WAI-IG list and the world in general, here are 4 implementations, try them via these proxies, and let us know which worked best for you and why, and what equipment you were using when you used them; then we build up empirical evidence AND make the tools/repair functionalities available to ALL users; UA WG shying away from repair functionality, so we need to (A) pick up their slack as regards the repair function of user agents, even if only by proxy or plugin, and (B) show implementers the way to add evaluation and repair functionality to AU tools WC: then all we have to document is what works, rather than theorize and then wait to see what works--I like that development cycle better than the present

CR: working in that direction now; problem is logistics-- many techniques not in tool due to either shortage of people to work on them, or complexity of issues involved

GJR: not denigrating current work, but pointing way forward that will expedite our work and provide tools for people to use today, and which actually repair on the fly, rather than limit our work to helping authors as they create pages

WC: to do group activity, have group use particular programming language and platform to assign action items and move forward

GJR: one thing we should concentrate upon is getting ER functionalities into Amaya, since that is test bed client for W3C; we need to liaison with them, and, probably, given their logistical limitations, do some of the work for them, or get some of Daniel's students to work on putting ER functionalities into Amaya

WC: what about CAST and A-Prompt people? what do you want out of the ER group? would you be ok if we worked with Amaya?

CR: from my perspective it would be OK; ATRC wants to get A- Prompt integrated into as many tools as possible, including Amaya

HB: would you want to do part of integration yourself

CR: would help with integration as much as possible; optimally working with someone on Amaya team,

WC: is A-Prompt open source?

CR: going to be

WC: Amaya is written in C, but uses different libraries for different platforms. so I believe that there is, essentially, a totally different code base for each platform;; code is currently a mess, as GJR said, due not to the efforts of the Amaya team, but because a lot of it was hacked up for someone's Ph.D. dissertation or MS thesis; Amaya does some interesting things; created a language to create graphical layout through libraries, which sounds like the XUL idea to me;

LK: how Windows-specific is A-Prompt?

CR: right now, it is only available for Windows--basically, A-Prompt is a Windows DLL that is called by another program

WC: could easily integrate with Windows version of Amaya

LK: that fist into the spirit of prototypes; if works in Windows, then figure out how to make work on other platforms

WC: how close or realistic is A-Prompt becoming open source?

CR: details have to be worked out; problem is that the project received funding from a number of sources, need to check with all of those sources to find out if any intellectual property rights claims are attached to any of them; if make A-Prompt open source, then could get a lot more people working on it, so I personally support making it open source

WC: if can resolve open source question before F2F, that would be a major piece of planning and strategy

CR: I'll look into it this week; Michael and Brian, please do the same at CAST

// ACTION CR: investigate A-Prompt's status vis a vis open source //

// ACTION MC/BM: investigate CAST's plans for making Bobby open source //

LK: even though it is implemented as a Windows DLL, to what extent does A-Prompt separate logic from repair?

CR: evaluation very portable; problem is how we "tokenize" HTML object; we've been talking with CAST about using Java; but that is probably the longer term solution-- need to stick to Windows now, so that we can deliver a robust and functional tool

LK: rule interpreter should have nothing to do with user interface

CR: rules implemented in code; isn't Bobby starting to implement rule base using XML?

MC: XML document not driving rules now, just providing language that gets pulled into report; planning on moving Bobby version 4 in that direction; I've actually been doing a lot of work on that, and it is a lot more complicated than it sounds

LK: sounds complicated to me!

WC: me too; we need an ERT need a schema, sort of like Schematron, but is the problem that not everything can be done logically? let me step back a bit--how many people familiar with RDF and Schemas?

LK: cursory idea of it; know XML and RDF conceptually

DB: have conceptual grasp, too

HB: know XML

MC: dream in XML but only know basics of RDF; looked at Schematron -- like ideas, but question is how much can be expressed logically in language used; there are some things that Schematron couldn't do that Bobby does very easily; why? my theory is that Schematron tries to use a meta language to abstract techniques--mentally, I've been thinking how far CAST could go down that path; need to think of ways of extending abstraction; RDF and Schemas relate to document, but for things that can't be expressed in terms of HTML structure--I can't see how would be addressable using Schemas or RDF

LK: like what?

MC: use clear and simple language; use consistent navigation- -if structure not consistent from page to page, for example, but the author uses consistent navigation, mechanisms, that may be problematic for a Schematron-type tool to deal with

WC: MC, would you post some examples and some thoughts as to how one to might express stuff like that to the list?

// Action MC: send example to list of what Bobby can do easily that Schematron can't do, as well as ideas for relating RDF and Schemas to those things that can't be expressed in terms of HTML structure //

WC: need to keep momentum; need to educate ourselves in some of the languages; could take time at F2F or telecon to do a half-hour presentation on a language or format and how it might work, what limitations are, etc.

// GJR suspends minuting due to PC problems;

// LK minutes while GJR's computer fritzes //

cr: you mean suggesting another form for ert

wc: no,not rewrite ert, but lets all have same language, xml, rdf? I'll investigate. michael said rdf not enough

lk: no just schematron

mc: rdf and schemas are structurally based. work well with structure of doc, like presence of alt, but can't express other stuff

lk: like what

mc: use clear language, use consistent navigation... structure not consistent even if navigation is

wc what about fuzzy logic? to combine test results, different opinions? fuzzy consensus

db: what's question?

mc: problems: structure, can express in rdf. but other issues, like determining language, no pattern to match. but can use fuzzy logic. can run number of tests and combine to come up with one answer. is that something we want to investigate? educate us? develop technique for it. How to express?

db: tricky area even though similar things are done. can refer to those that are available. but so much more in web pages. very tricky.

wc: must first define problem. are several methods to determine reading level rating... syllables, words used, words in sentence... mostly straightforward

mc: concern is how valid is the rating

wc: educators have different levels. but can combine in answer. but not just one of tests.

db: checkpoint about languge is not main point. usability with assitive technology, tables, style sheets causing most problems. I think this doesn't need as much work.

gregory: can do screenscrape and analyze

mc: but do people want that? there are lots of lynx me services .

gregory: usability belongs more properly in eo. may have someone who can help with ai and fuzzy logic.

wc: we should be looking at harder issues. ai, fuzzy logic, if they can help lets use. we've been avoiding hard things. maybe expert system,

db: seem straightforward but no tools that can help. can be hard to do.

gregory: so use proxies to get feedback.

wc: e.g. low hanging fruit... implement to show how to do. do easy things first. where to begin?

mc: that's why atrc and cast have just started doing things simply and well. Then move on. do easy things first. helps to focus and know you are getting somewhere.

wc: have you filled checklist for bobby

mc: not yet

wc: do it. will help set priorities.

// GJR resumes minuting in earnest //

WC: should be looking at higher end issues, yes--if can do with "fuzzy logic" or AI, or whatever

DB: some things seem straightforward, but not easily tested

WC: need to go through P1 checkpoints, find out what has been implement, and what hasn't--CR did you post a note alerting the group that the checklist has been posted?

CR: yes

// LK, GJR seen it //

[Scribe's Note: The AERT Checklist can be found at: <http://www.w3.org/WAI/ER/IG/ert/ert10-chklist.html>]

WC: where do we start?

MC: CAST is concentrating upon doing what we can do simply and do well, once we've done that, we plan on moving on to address the harder issues; there are many things that Bobby could do, but we felt that we need to do the easy things first; helps to focus things to know you are getting somewhere and have a tool that actually does something

WC: have you filled out the ERT checklist for Bobby?

MC: can, but haven't submitted it

WC: please do; the more people/developers who fill out the checklist, the quicker and more accurately we can set priorities as to where do we go from here? need to concretely go through checklist and fill up holes; have people submit ideas and move forward; need to assign action items

CR: are you going to look at the ATRC document in a prose format to look at evaluation without having to understand technology? advantage to having prose is that one need not be fluent with the technology to understand the problems

LK: we're not saying "change it" WC: right--we want to fill in gaps and see things implemented

CR: are there any programmers here with time on their hands that could help with coding for A-Prompt

WC: ME!!! I'm ready to do coding big time--send me something to chew on

CR: will do

// ACTION CR: send WC something to chew on //

LK: I have long list of functionalities to add to WAVE

WC: Daniel has students coming to work over the summer, but need to give them something to work on; CMN could be liaison between us and Amaya

// ACTION LK: put feature into WAVE that will allow someone to look at source to see if what appears to be a list is a list; add logical tags so that if there is a numerical list outputs an icon //

WC: favorite feature of HoTMetaL is graphical tag view

LK: Composer has it too--tag symbol with name of tags; in spirit of making accessible to HTML-aware people, use specific tags to represent specific elements

// DB leaves //

LK: want to beef up WAVE for F2F

GJR: would like to work with LK on making output generated by CSS pseudo-elements rather than icons

// WC experiences phone problems; her phone may die soon //

WC: other agenda items?

LK: there is an "if time" agenda item dealing with the Minimum Requirements for WCAG 2.0 document, but I'd prefer to keep focus on ER right now

WC: at WCAG F2F people said checkpoints need to be easily verifiable

LK: would remove word "easily"--if hard to do, that's the way the cookie crumbles

WC: ok--clear is a better word

LK: right--goal is that 2 people can look at page and unambiguously say passes or doesn't

WC: if WCAG can do that, it would be a huge win

LK: comments in regards guidelines is making them precise enough

LK: any other issues? there's no law that says we have to go to 11:30

WC: I'm fine with ending

CR: are we meeting next week?

LK: good question--let's see; we may have last minute things to discuss

WC: at least report on action items

LK: exchange notes on who is coming and not; action item updates

WC: date next Monday--the eighth -- day registration closes

LK: how public is access to registration list?

WC: list of registrants?

LK: yes

WC: don't know if member private or team private; can comment on it next week aurally--give you a list

// ACTION: GJR and LK send minutes to WC for encoding and posting //


$Date: 2000/11/08 08:17:43 $ Wendy Chisholma