- From: Wendy A Chisholm <wendy@w3.org>
- Date: Thu, 07 Feb 2002 18:50:03 -0500
- To: w3c-wai-er-ig@w3.org
<wendy> nick - is there still a way to get static output from page valet? <wendy> while stepping through the errors is cool, there's not a good way to skim through them. <wendy> i thought the "document view" link would give me that, but it just described the different views. [18:48] <nick_kew> hi wendy. [18:48] <nick_kew> just a mo [18:49] <nick_kew> Use the "messages+source" view - though I may have broken it yesterday <wendy> hmmm. tried that. i'll try again. [18:51] <nick_kew> It's under development; not really satisfactory even if I didn't break it <wendy> trying again...get the page view and have to tab through messages. [18:55] <nick_kew> OK, yes, I did break it [18:55] <nick_kew> Try again now [18:57] <nick_kew> (the "highlight" links require javascript to work well - without it they just go to the element but don't highlight it) [19:05] <nick_kew> sbp, have you seen FillyJonk? <wendy> hmmm. i have javascript, but nothing is highlighted.... [19:06] <nick_kew> It might be out of the window you can see. Click a second time <wendy> you've taken the WCAG errors out of the main source and put them along the left side? [19:06] <nick_kew> Yep <wendy> ah. i see what's happening. why move them out? i liked them in w/the other errors. :) found it much easier to read. [19:07] *** JibberJim (~Jim@bh-cw32-152.pool.dircon.co.uk) has joined #er [19:07] <nick_kew> Well, the default document view puts them inline [19:08] <JibberJim> Hi, apologies for being late - my dodgy cable! [19:08] <nick_kew> hi <wendy> ah - ok. was that broken before also? i thought i tried that. [19:09] <nick_kew> No, that was fine. That's always been the "primary" mode of use <wendy> I still find the WCAG1, WCAG2 confusing. [19:09] <nick_kew> In the messages? [19:10] <sbp> <nick_kew> sbp, have you seen FillyJonk? [19:10] <sbp> of course - I was in #validator when you guys were testing it <wendy> I see the issue w/expressing confidence as "WCAG1/High" or "WACG2/Low" Seems to be 2 priority levels not one for priority and one for confidence. [19:10] <nick_kew> Ah, of course... <wendy> nick - yes WCAG1, WCAG2 in the messages. It makes it seem like WCAG version number rather than priority level. <wendy> we all seem to be here...already discussing Nick's questions...continue on with that? [19:11] <JibberJim> I want some advice from you sean on how to present the result of EARL queries in an understanable form in IRC, then I could add in some EARL querying to FillyJonk (it's valuable as it's all javascript I'd re-use in other scripts.) [19:11] <nick_kew> Maybe I can change them to WCAG:A, etc? [19:13] <sbp> Jim: Hmm... I'd represent them as N3 of course, which was designed for scribbling down RDF in environments such as this [19:14] <JibberJim> How easy is triples to N3 ? [19:14] <sbp> but OTOH I use N3 for everything, so you have to take that cum grano salis [19:14] <sbp> triples to N3: very easy [19:14] <sbp> if you can spew them out as XML RDF, you can surely do it as N3 [19:15] <JibberJim> I'll look at that then tomorrow... [19:15] <nick_kew> Wendy, is it any better now? [19:17] <JibberJim> Do we have an Agenda? <wendy> yes agenda posted to the list today. * wendy digs for uri <wendy> http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Feb/0019.html [19:18] <JibberJim> I was too busy moaning about the "tonight" :-) <wendy> nick - it's interesting to use A, AA...I understand it. wonder how other people will do with it. <wendy> we've discussed some of nick's questions. <wendy> shall we continue on with that then get to Jim's? <wendy> I sent the agenda to Marja, EriP, and Ralph w/the hopes that one of them might show up for the chat. [19:19] *** Masaya (~ando@210.255.84.226) has joined #er <wendy> no luck, it seems. although, i did chat for a while yesterday w/marja and i'll be at MIT in a couple weeks and we'll work more on this stuff. <wendy> Welcome Masaya! [19:19] <nick_kew> Wendy, my main question now is to discuss how well these confidences work [19:19] <Masaya> Hello [19:19] <nick_kew> hi <wendy> Everyone - Masaya is working on the J-WAS project - an evaluation tool in Japanese. [19:20] <Masaya> I am Masaya from Tokyo <wendy> They are considering implementing EARL. <wendy> Masaya - we are discussing Nick's tool called PageValet at: [19:20] <JibberJim> That's good the more developers the better. [19:20] <JibberJim> Hello Masaya. <wendy> http://valet.webthing.com/page/ [19:20] <Masaya> hi [19:21] <Masaya> ok... I will see... <wendy> Nick - you are wondering about the confidences - you mean e.g. "WCAG-A:High" or "WCAG-AA:Low"? [19:21] <nick_kew> Well, two things [19:22] <nick_kew> firstly the High/Low/etc values [19:22] <nick_kew> secondly how best to express them (needs an explanation page, too) [19:23] * sbp waves to Masaya <wendy> with WCAG-A/Low you are saying it passes with low confidence or fails w/low confidence? [19:23] <JibberJim> I'd really like numbers rather than high low etc. <wendy> oh, I guess fail, eh? [19:24] <nick_kew> Yes. All the messages are saying "possible fail" <wendy> Can you be more explicit in the message? something like: Fails but needs human testing - Priority 1 WCAG checkpoint: blah blah blah" [19:24] <sbp> the inclusion of "-" is interesting; perfectly uncertain fail == pass [19:25] <nick_kew> sbp, yes, that's because there were two null tests in Section508, and I put in the messages for legalistic completeness [19:25] <nick_kew> (i.e. so as to test the checkpoint) <wendy> or: Possible failure - please confirm. WCAG Priority 1 checkpoint blah blah blah... [19:26] <nick_kew> wendy, I'm trying to avoid verbosity. These reports are already long! <wendy> 7 extra words is too long eh? [19:27] <nick_kew> I've seen pages generate >500 messages. 500*7 is a lot <wendy> what about: Fails, Possibly fails, Uncertain failure - or something like that? <wendy> as 3 diff levels? [19:28] <nick_kew> Well, [19:28] <nick_kew> I think the ideal thing would be a visual cue in CSS [19:29] <nick_kew> the words are just a fallback <wendy> what if you give someone a choice of viz cue, text, viz cue+text? <wendy> (3 separate choices) <wendy> jim - you prefer numbers? what kind of numbers? [19:30] <JibberJim> it's XML (or RDF) - we have a choice already... [19:30] <nick_kew> Ug - lots of stylesheets to maintain [19:31] <JibberJim> for confidence of EARL predictions, I'd prefer it if we could use numbers rather than just high low... <wendy> e.g.? [19:31] <nick_kew> Jim, I'm already worried about my assignments. With percentages it becomes even harder [19:32] <JibberJim> valet says pass confidence 60% - Snork says pass confidence 70% - then we can combine the two easily and with appropriate rating on how well the different tools assess that checkpoint. <wendy> re:css - then i think the default ought to be viz cue+text <wendy> how do you determine the percentage? [19:33] <nick_kew> Jim, you need an extra link in there: how much do you weight each of two tools [19:33] <JibberJim> That depends on the tool, for high/low/definate we can just have 100/66/33 or whatever. <wendy> e.g. provide an accessible alternative to script elements. if there isn't noscript, then what percentage confidence can you provide? [19:33] <JibberJim> that's a user issue Nick depending on how you use the two reports. [19:34] <JibberJim> I'd say that's high for code not authored by about 3 people in the world, low for the other 3... [19:34] <nick_kew> wendy, I take that as low confidence because <script> often has no accessibility problems, even without noscript (it may not be doing anything critical in the first place) <wendy> do we really need more than 2 levels? e.g. can we just say "passes" or "fails"? [19:35] <nick_kew> No, because Valet can't be sure in many cases [19:35] <nick_kew> e.g. the script on Jim's page <wendy> right. so passes, fails, not sure. [19:36] <nick_kew> But I'm using a finer-grained notsure. <wendy> do you need to? what benefit does the user get? <wendy> if i have a list of things I know failed, i'll focus on those. <wendy> if i have a list of things that might have failed with low certainty, i might ignore. <wendy> if i have a list of things that might have failed with high certainty, does that say much more to me than those w/low certainty? [19:37] <JibberJim> I (and many users) would rebel against tools which told me my accessible (by human judegement) page fails. [19:38] <nick_kew> I think it's important that the user can see which messages mean "that's probably wrong", and which ones mean "probably OK, but you might want to check" [19:38] <nick_kew> Jim, do you think Valet's "probably ..." approach will work? <wendy> jim - i'm not saying that the page should say "you fail" i absolutely agree we need the human to check and make the final decision. [19:39] <JibberJim> For humans in isolation yes, for combining I'd convert them to numbers in any case. <wendy> will you have an interface that let's the user make assertions? e.g. valet says this one might fail, but i've tested and determined it doesn't. [19:40] <nick_kew> Also it's not saying this is[n't] accessible; it's saying this does[n't] pass some test suite <wendy> then valet could update the report? [19:40] <nick_kew> Wendy, yes, but that's some time away [19:40] <JibberJim> I think that's related to my agenda item in any case... [19:41] <nick_kew> That's where EARL becomes more useful ... <wendy> exactly - one of the main reasons for it. [19:41] <nick_kew> HTML includes <link rel="accessibility" href="earlstuff.rdf"> <wendy> so, nick, whatcha gonna do? :) [19:42] <nick_kew> Valet reads assertions, and applies them. And provides an interface for the user to generate the assertions [19:42] <nick_kew> But .. lots of work before it does all that [19:43] <nick_kew> Jim - your agenda item: Are you ahead of me on this one? [19:43] <JibberJim> I would hope so, but I need some help :-) [19:44] <nick_kew> From me? [19:45] <JibberJim> Nope, From someone who knows about RDF databases and has a server to run them on... [19:45] <JibberJim> That is unless Annotea is already a solution. [19:45] <nick_kew> [mumble] .. another thing on the neverending todo list [19:46] <nick_kew> wendy, can you find an annotea expert to tell us what it can do for us? <wendy> i tried to get marja, ralph, or eric here today... <wendy> i was able to get annotations from jim's page using amaya the other day. very cool. [19:47] <nick_kew> I can't get amaya to work :-( <wendy> at all? <wendy> that's surprising. [19:48] <nick_kew> Well, it can view one or two pages before crashing <wendy> what version? [19:48] <JibberJim> Eric describes a general RDF database at http://www.w3.org/2001/Talks/0505-perl-RDF-lib/ which I assume could be used to store EARL easily, however the examples aren't clear. <wendy> re: rdf databases, sean, isn't libby a person we could talk with about that? [19:48] <sbp> indeed <wendy> jim - is your issue that you want to store only rdf and not w/any annotea wrapper? [19:49] <sbp> and Sandro et al. All of the rdf-rules community [19:49] <nick_kew> amaya 5.2 <wendy> do you know if any of them is around to pull into our chat right now? or perhaps i ought to schedule a time for them for next week... [19:49] <JibberJim> The Annotea wrapper seems to make it useless - I can't make any queries on it. <wendy> why not? [19:50] <JibberJim> Because Algae which use to query the annotea - doesn't look inside the annotation. <wendy> ah - thus why i was downloading the annotation/earl to view rather than amaya displaying it to me? [19:50] <JibberJim> That's related certainly. <wendy> so, is that a problem with the algae implementation or the annotea idea? <wendy> if it's the imp, then talk with ericp. if not, try to set up our own rdf database? <wendy> i think an earl-only database would be good for several reasons... [19:51] <JibberJim> I think the annotea idea (with respect to EARL), but I don't know for sure that it couldn't be changed - I prefer the EARL own though, and see that as a easier solution in any case. <wendy> primarily so that clients wouldnt have to speak "annotations" [19:51] <nick_kew> I downloaded an rdf-on-mysql thingey - possibly from annotea <wendy> athough if they did, we could combine data from both. <wendy> are any of you planning to attend the w3c technical plenary? [19:52] <JibberJim> We could have our EARL RDF database store Annotations aswell, they are just RDF after all - it would just need more intelligent clients. [19:53] <nick_kew> On the TBD list is to see if I can port it to pgsql so I can run it with valet <wendy> is the annotations database just rdf? if so...couldn't we store earl there? <wendy> seemed to me that eric was saying something about a very generic store <wendy> so that no matter how the annotations evolved they could store anything there. [19:54] <nick_kew> I think it claims to be just rdf [19:54] <JibberJim> I believe it can yes, but if you chuck EARL at it directly it just complains. <wendy> sounds like we ought to talk w/eric p. [19:54] <JibberJim> the Annotea server doesn't understand anything but Annotea, but underneath the database does. [19:55] <nick_kew> Jim, what's your database at the moment? [19:55] <JibberJim> For the old earl client? [19:55] <nick_kew> yep [19:55] <JibberJim> A sort of hacked together mysql tables. [19:56] <JibberJim> It's not good enough and needs replacing. <wendy> http://www.w3.org/1999/02/26-modules/User/RdfSQL-HOWTO.html <wendy> "The Annotations server is a narrow shim on top of the RDF database. <wendy> " [19:57] <nick_kew> http://www.w3.org/1999/02/26-modules/User/RdfSQL-HOWTO.html [19:58] <nick_kew> oh, .. blush [19:58] <nick_kew> That's the one I'd like to port to pgsql <wendy> ah. [19:59] <JibberJim> I should really be able to get it going, but I dislike perl so would have to port the "shim"... <wendy> instead of perl, what would you use? [20:00] <JibberJim> I also don't really have a server for it :-( [20:00] <JibberJim> javascript of course... <wendy> check out: http://web1.guha.com/rdfdb/ <wendy> uses perl...although someone did a port to java...how do you like java? [20:02] <JibberJim> I prefer perl wendy... <wendy> jim's order of pref: javascript, perl, java [20:02] <JibberJim> with about a million dots between perl and java [20:03] <JibberJim> porting the shim is probably not to hard, I could even use perl for any difficult bits. [20:03] <JibberJim> I couldn't make the server widely available currently though through lack of bandwidth resourced. [20:03] <JibberJim> What do people want from an EARL database in any case? [20:04] *** SeanP (~sean@m48-mp1-cvx4c.pop.ntl.com) has joined #er <wendy> check out all the apps: http://www.ilrt.bristol.ac.uk/discovery/rdf/resources/#sec-apps [20:05] <nick_kew> Well, the kind of usage I was describing earlier will be for intranet users. Too heavy to put up in public, esppecially with the overhead of securing it <wendy> rdf query in javascript: http://www.w3.org/1999/11/11-WWWProposal/rdfqdemo.html [20:06] <JibberJim> that's been kind of superceded by http://jibbering.com/earl/rdfquery4.html - but mine doesn't have documentation. <wendy> cool [20:07] *** sbp has quit IRC (Ping timeout) <wendy> well, the point is...there are lots of resources for creating rdf databases. <wendy> but, you ask a good point - what do we want to do w/it. [20:07] *** SeanP is now known as sbp <wendy> what i want: [20:08] <nick_kew> but how mature/robust are they? <wendy> a place where multiple tools can store and/or retrieve earl data. <wendy> whether it be for a search, a repair, or an evaluation. <wendy> then query it depending on the app. [20:08] <JibberJim> I think we should research them Nick, and move on to what to do with it. [20:09] <JibberJim> Do you log evaluations Nick? could you easily send all to a database - either live or in batch mode? [20:10] <nick_kew> The only evaluations that go into a database are the manual ones (the user feedback form) [20:10] <nick_kew> - and those from the site monitoring service [20:11] <nick_kew> NOT from the developer tools like Page Valet <wendy> could they? (the ones from page valet go to a db)? [20:12] <nick_kew> Yes. But I don't have the resources to do that on my poor little server [20:12] <JibberJim> Than should be easy to create Wendy, all we need is a server, if we could get Eric interested in creating I'd've thought the server could be up in very little time - then I could easily build some Mozilla/IE based clients. <wendy> cool. [20:13] <JibberJim> Can you get Eric on board for that? <wendy> i'll try, yes. [20:13] <JibberJim> How about a server for it - is that feasible? [20:14] *** SeanP (~sean@m4-mp1-cvx4c.pop.ntl.com) has joined #er [20:14] <nick_kew> [we assume W3C has more server horsepower than we do] [20:15] * JibberJim is more concerned about the bandwidth for the ones at his disposal [20:15] <nick_kew> - that too:-) [20:15] <nick_kew> Mind you, I could upgrade my server and bandwidth easily, if someone paid [20:17] <JibberJim> Is there anything in Seans 1.0doc ? didn't you have a new "suspectsomethingorother" in valet Nick? [20:18] <nick_kew> That was before I implemented confidence levels [20:19] <nick_kew> (btw, sbp's spec defines High/Medium/Low for them) [20:19] *** sbp has quit IRC (Ping timeout) [20:20] *** SeanP is now known as sbp [20:20] <sbp> [[[ [20:20] <sbp> earl:ConfidenceLevel a rdfs:Class; rdfs:label "ConfidenceLevel"; [20:20] <sbp> is rdf:type of earl:Certain, earl:Likely, earl:Unlikely . [20:20] <sbp> ]]] [20:20] <sbp> - http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Feb/att-0007/01-1.0.n3 [20:20] <JibberJim> Sean seems to be having problems.. [20:20] <sbp> indeed [20:20] <nick_kew> I think [high|medium|low] come from an earlier draft [20:20] <sbp> yes, from 0.95 [20:21] <JibberJim> Can we not have a choice of types? [20:21] <sbp> is there any consensus on the range of levels needed? Jim asks for a numberic value, Nick now uses an extend range... [20:21] <sbp> choice: of course. it's not a strict enumeration [20:22] <JibberJim> so RDF:type means what - it could be this, but? [20:22] <nick_kew> Numeric (eg percentage) values would be the most flexible [20:23] <nick_kew> but I'd just be using them as synonyms for high/low/etc [20:23] <sbp> the earl:confidence links a validity property to its confidence i EARL [20:25] <nick_kew> Jim, would percentages suit you, even if they were just used as synonyms for something more fuzzy? [20:25] <JibberJim> Of course, I don't mind how some tool gets to the % [20:26] <JibberJim> I just would like when I'm combining 3 evaluations combine them with different weights and thus make a more true comparison. <wendy> then why don't we use weights? 1=high, 2=medium, 3=low? <wendy> how would you combine them? [20:26] <nick_kew> Right. But the way I'm working, that translates to <wendy> if report one=1, report 2=3, and report 2=3 - do you take the average? [20:27] <nick_kew> Wendy, that's exactly what I do now: it's just that the weights have names <wendy> get a confidence level of 2.3 [20:28] <nick_kew> enum { Test::Low = 1, Test::Medium=2 , ... } confLevel ; [20:28] <JibberJim> An average yes if all tools were equal but if one was Bobby and one was page valet, I'd want to give page valet a higher weight, and if one was human I'd do even more) this could mean I could get a rating of 1.5 where would that go? [20:28] <nick_kew> could scale those up and call them percentages <wendy> i'd like to see these clearly defined. <wendy> e.g. when does a low-high become a medium? <wendy> e.g. 2.3 - is that a medium? or low? [20:29] <nick_kew> Use Bayesian or neo-bayesian logic [20:29] <JibberJim> Exactly so if we have %'s we don't need to worry we just give them the % and the tool decides... <wendy> i want some clean, testable rules! :) [20:29] <nick_kew> (I'm using something more akin to a Dempster-Shafer belief function) <wendy> what if the human wants to decide? e.g. - 3 tools said you pass and one said you might fail. <wendy> instead of "you pass with 70% certainty" <wendy> although, that could be helpful, too. [20:30] <nick_kew> Jim, have you worked much with frameworks like this that combine evidences? [20:31] <JibberJim> No. [20:31] <JibberJim> I did a little a long time ago, but then I was never paying attention at Uni in any case... [20:31] <nick_kew> I have, and it's harder than it looks to get sensible results out [20:32] <JibberJim> I'd still like it to be possible, but can always implement it over the top. [20:32] <JibberJim> high/low etc. do map to numbers without problem. [20:32] <nick_kew> (when I did speech recognition research we had an evidence-combination engine, but it never outperformed an old-fashioned neural net) [20:33] <nick_kew> - and the research money ran out just as we were in a position to start getting some really interesting results * wendy searches for rdf projects that combine results... [20:34] <nick_kew> OK, this is a much easier problem than speech recognition, but nevertheless... [20:36] <nick_kew> Jim, I'm kind-of thinking a human evaluator will be allowed to assert things with 100% certainty and overrule anything a tool says [20:36] <JibberJim> Of course! <wendy> i agree. <wendy> although, what about some people who might not feel comfortable and want a second opinion? <wendy> i don't think we should rule out uncertainty by humans. [20:37] <nick_kew> And that brings us to #2: what happens when the [human+tool(s)} gets it wrong? [20:38] <nick_kew> - the need for a feedback system, where other humans can submit well-structured problem reports <wendy> i wonder how amazon and the IMDB rate user feedback? <wendy> ah duh, the user submits a subjective rating on a scale. [20:38] <nick_kew> I mean user feedback dealing specifically with accessibility <wendy> so, maybe some sort of scale is the best way... [20:39] <nick_kew> wendy, yes, humans can assert uncertain things too! [20:39] <sbp> { ?x earl:asserts [ rdf:subject ?y; rdf:predicate [ earl:confidence [ earl:pc ?pc1 ] ]; rdf:object ?z ] . [20:39] <sbp> ?p earl:asserts [ rdf:subject ?y; rdf:predicate [ earl:confidence [ earl:pc ?pc2 ] ]; rdf:object ?z ] . [20:39] <sbp> ?result math:quotient ([ math:sumOf (?pc1 ?pc2) ] "2") } log:implies [20:39] <sbp> { this earl:asserts [ rdf:subject ?y; rdf:predicate [ :averageConfidence ?result ]; rdf:object ?z } . <wendy> sbp - where in the rdf spec does it talk about combining results? [20:40] <sbp> in the RDF spec? the RDF spec talks about graphs, and what happens when you merge them. you just combine the graphs [20:40] <nick_kew> and then that assertion would NOT override the tool, though it will of course modify the overall belief state [20:41] <sbp> [conjunctively] [20:42] <nick_kew> RDF can say two arcs go between the same nodes (different assertors assert assertions on the same subject+object) [20:42] *** JJ (~Jim@bh-cw31-059.pool.dircon.co.uk) has joined #er [20:42] <nick_kew> but it's up to the tool to reduce that to a single {arc|assertion} and combine confidence levels [20:43] <sbp> indeed [20:44] <sbp> s/quotient/quotientOf/ <wendy> sbp - mind if i do some editing on the EARL 1.0 spec? [20:46] <sbp> of course not - please go ahead [20:46] *** JibberJim has quit IRC (Ping timeout) [20:46] *** JJ is now known as JibberJim <wendy> is the recent EARL document that you sent just to test out your ideas for modifying earl 1.0? <wendy> i.e. based on previous suggestions? [20:47] <sbp> yes - it [20:47] <sbp> argh [20:47] <nick_kew> pardon? [20:48] <sbp> it's just edited from 0.95, incorporating soem of the thigns that have been raised on the list, and in creating tools [20:48] <sbp> my ' key is too close to \n [20:48] <nick_kew> hehe [20:48] <JibberJim> Arghh you need an Azerty keyboard then... [20:49] <JibberJim> not because the ' is anywhere different, it'll just slow you down so much you won't make mmistakes... [20:49] <sbp> I'll just prise the keys out, and rearrange them... [20:49] <nick_kew> Chicken and egg... [20:49] <sbp> I could use Dvorak [20:49] <nick_kew> We need to create tools to get experience of what EARL wants ... <wendy> ideally - we need an implmentation of the "end to end" process discussed at the october F2F. [20:50] <nick_kew> We want a stable-ish EARL spec to create tools [20:50] <JibberJim> Give me a database, help me fill it, and I'll give you at least 1 and 1/2 tools. [20:50] <sbp> yeah, the chicken/egg dilemma is rather annoying <wendy> i've sent a note to eric p re: database. <wendy> see the ert wg list for the msg, i cc'ed y'all. <wendy> i'll be at MIT in a couple weeks to work with Marja, Ralph, et al. [20:50] <JibberJim> I saw Wendy... [20:51] <nick_kew> Wendy, if we have that database + my server tools + Jim's client stuff + my problem tracking database... [20:51] <JibberJim> Thanks! <wendy> right. and hopefully we'll get a couple other implementations. [20:52] <nick_kew> other implementations? Why aren't the implementors on #er? [20:52] <JibberJim> I think if we can demonstrate some good use scenarios people will come. <wendy> i've been trying to get them here. and, jim's right, the more imps we have the easier it will be for me to get them involved and convinced. [20:52] <nick_kew> I thought wendy meant others working on it now <wendy> there are 3 possibilties. one got the go ahead and we should see a new imp in a matter of months. <wendy> the other 2 are still not sure. [20:54] *** Masaya (~ando@210.255.84.226) has left #er (Masaya) <wendy> QA will be using EARL in test suites. [20:54] <nick_kew> test suites for??? <wendy> CMN has a new student working on this tuff. <wendy> w3c test suites for specs. [20:55] <sbp> Whatever happened to TPDL? [20:55] <nick_kew> Would that be relevant to me? [20:55] <JibberJim> CMN should get his student in here... [20:55] <nick_kew> CMN should get himself in here [20:55] <JibberJim> true... [20:55] <JibberJim> but it's easier for him to get his student in... [20:56] <nick_kew> [aside: is English less-than accessible to some who would like to participate?] <wendy> cmn requested that we meet at a time that's not in the middle of the night to get his student (nadia) here. <wendy> i also suggested she show up during the day, as we come and go. <wendy> TPDL - being discussed in QA. [20:57] <nick_kew> wottawimp:^}! [20:57] * JibberJim would like that too - not for the time, just as I'd have a proper pc/net connection. <wendy> what about next wednesday we meet at...hmmm, a time that will be ok for Masaya as well... [20:58] <nick_kew> Nick prefers after 6pm GMT - for cost reasons [20:59] <JibberJim> Do you not have an unlimited home connection Nick? <wendy> 6+9=3 a.m. in Tokyo? [20:59] <nick_kew> Not as of now. Keep meaning to find an ISP with a decent 24hour package that works [20:59] <nick_kew> (any suggestions?) [21:00] * JibberJim doesn't have one, only connects for email and this outside the office. <wendy> maybe next week we could meet for an hour at 6 GMT, to chat w/Nadia and MIT folks, then again at midnight GMT to catch Masaya? <wendy> or we rotate between these times each week? <wendy> although, not sure that 7 p.m. is reasonable for Nadia. [21:01] <sbp> we should be having 3 hour meetings every day at 5AM, 6PM, and 12AM <wendy> Nick - sometimes you connect during the day... <wendy> lol [21:02] <nick_kew> Wendy, only when I have some very particular reason <wendy> ok. well i'll check w/nadia and try to schedule something that works for everyone. i'll also try to get eric or someone here. <wendy> i think ideally, if we could meet around 4 GMT that would be good, but I don't want to hurt Nick's pocketbook. [21:03] <sbp> we can bring in Aaron and xena to pad things out a bit too. Is there any chance of Chaals getting in here? <wendy> chaals - perhaps if it is during the working day hours in Nice. -- wendy a chisholm world wide web consortium web accessibility initiative seattle, wa usa /--
Received on Thursday, 7 February 2002 18:45:53 UTC