ERT face to face - day 2

18 Oct 2005

See also: IRC log


Shadi, CarlosV, Nick, Yehya, Charles, Jim, CarlosI, Sandor, Johannes
Johannes, Nick, Charles, Jim




<niq> http://lists.w3.org/Archives/Public/www-annotation/2002JanJun/0156.html


<JohannesK> NK: URI and XPointer are not sufficient for several problems

<JohannesK> NK: e.g. misspelling of a certain word in some paragraph in some document at a given time

<JohannesK> NK: reduce markup to some normalized version in order to detect different levels of change

<JohannesK> NK: also works with some (not too bad) tag soup

<JohannesK> NK/SAZ: some test results may be affected by certain changes, others may not

<niq> http://lists.w3.org/Archives/Public/www-annotation/2002JanJun/0076.html

<JohannesK> SAZ: do we have to double-check everthing?

<JohannesK> JL: not necessarily

<niq> http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Feb/0002.html

<JohannesK> JK: there is a difference between automatic and manual tests here

<niq> http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Jul/att-0017/metrics

<shadi> http://lists.w3.org/Archives/Public/public-wai-ert/2005Mar/att-0062/errors

<JohannesK> JL: it's not only about performing the test, but also displaying test results

<niq> <hash method="my-method">value</hash>

<JohannesK> JL: there are more types of metadata than hashes only

<niq> <invariance-measure method="my-method">value</hash>

<JohannesK> SAZ: persistency seems to have no top priority

<JohannesK> NK: provide a web service to calculate the invariance-measure

<JohannesK> SAZ: put hooks for persistence into EARL?

<JohannesK> NK: needs to be optional

<JohannesK> NK: per-page measures (HTTP headers: last-modified, content-md5, ...), per-element measures (hash on element structure)

<JohannesK> NK: put hooks into Assertion

<JohannesK> CMN: put hooks into Location

<JohannesK> JL/CMN: property of TestSubject

<JohannesK> SAZ: XPath may also be used

<JohannesK> SAZ: hooks for delivery unit level

<JohannesK> SAZ: do we need more stuff in WebContent?

<JohannesK> <http://lists.w3.org/Archives/Public/public-wai-ert/2005Jun/0102.html>

<niq> jl, nk: http headers should be stored parsed and queryable, not as a blob

<JohannesK> <http://lists.w3.org/Archives/Public/public-wai-ert/2005Jul/0014.html>

<niq> nk: can store http as <header> with name and value properties or as <name>value</name>

<niq> cmn: can use annotea's unofficial namespace for http headers

<JibberJim> http://www.w3.org/1999/xx/http

<chaals> ericP, question about http://www.w3.org/1999/xx/http - do you think we could take over that namespace and extend it a little?

<ericP> that was quick

<ericP> have to run off. will think about it...

<ericP> JibberJim, please relay to chaals when he finds his way back

<shadi> eric, you are being projected :)

<shadi> chaals can read you...

<shadi> we are at f2f

<niq> nk: don't enumerate headers because it makes it messy to extend

<niq> cmn: do enumerate http headers

<chaals> chaals: It is a bad idea to use things we don't enumerate.

<niq> saz: do we need both content-type and dc:format?

<shadi> http:ContentType

<niq> nk: probably good to have both

<shadi> http:lastModified

<shadi> http:language

<niq> jl, cmn: dc:format not very useful

<shadi> MD5, e-tag

<shadi> http:ContentEncoding

<shadi> http:ContentLocation

<niq> vary

<niq> content-*

<niq> accept-foo

<niq> cookies and authentication

<niq> cookies and authentication - may be problematic to store

<niq> saz: do we range for http?

<niq> cmn: probably no

<niq> ACTION: jk to update proposal for recording http transaction info [recorded in http://www.w3.org/2005/10/18-er-minutes.html#action01]

cmn: can store webcontent as body on that

nk: if we provide that hook, should also allow choice of representations

jl: people can use own has-copy; public copies are not earl's business

<chaals> [we could live without doing the work of defining what to use. dc:hasVersion is what I was thinking of]

saz: outstanding issue is invariance measure; subject otherwise closed
... we should sort the world's problems before word-smithing about webcontent further
... are we done with location once we have hooked in invariance measure?

"a rose by any other name ..."

<shadi> current proposal: <hash mathod="bla">value</hash>

<hash> <method>foo</method> < value > </>

<shadi> class: Representation

<shadi> properties: method, value

<shadi> (possibly further hasVersion recursions)

ericP: far too late, sorry

<ericP> chaals, what's the objective in taking over xx/http instead of extending it in another namespace?

<ericP> oh well

<chaals> [slightly reducing the number of namespaces people deal with]


saz: lunch!

SAZ: Request for everyone to have a look andd think about what other milestones we need to accomplish before we can get to last call

Nick: Would be nice to be able to describe actions and processes

SAZ: want to write that up?

JL: What is needed? Surely it is just a dscription of a test subject - which is outside of our scope of work. There isn't anything else the EARL needs to supoprt that

NK: None of our location stuff would handle it

JL: Right, but we are not limited to the ones we have - we just create some for our own use and EARL allows you tp create others

NK: agree with Jim.


JL: We should look at other use cases, and do smething if EARL makes it impossible to do, but we ddon't need to do all the work for ourselves.

SEP == Someone Else's Problem

SAZ: There are a couple of action items open, some changes to be made, so we can put together a new working draft mid-late november? Depends on how fast things are done.
... Next thing is the EARL Guide

SAZ: Anyone interested?

CarlosV: yes... what's the timeline?

-> http://www.w3.org/2001/sw/Europe/talks/200311-earl/all earl by example - step by step guide that should be updated and could be a basis

<scribe> ACTION: CarlosV get some shape to an EARL guide [recorded in http://www.w3.org/2005/10/18-er-minutes.html#action02]

SAZ: Next meeting 8 November

Chaals: Should we have the optional properties listed in the Schema?

SAZ: There are a lot of otional things. Developers can look at the schema and see at a glance what they need to support.

This makes the whole thing verbose, and is why there is the BNF, and so on.

JL: The developers who aren't going to read the prose are more likely to look at example EARL files than the actual schema, especially if the scchema is very verbose

CMN: Think the table is a lot clearer than mixing stuff into the schema. I like the table - but we also need to look at conformance requirements in general.

CV: Would like more examples, too
... and code examples using it would be good.

SAZ: Don't think it is a good thing to do stuff in a Rec based on specific implementation

CI: Would like examples, too

<JibberJim> chaals: lots of examples would be good, and hera can generate lots of those with ease

<velasco> Chaals said HERA has bugs

chaals: But does have to be updated to the latest schema version

CV: Since we are using DC and FOAF, some kind of appendix about those would be good

Chaals: Guide should cover that material

PROPOSED RESOLUTION: Optional properties will be dropped from the shema

SAZ: What if I want to get all the optional properties using Jena, to determine whether I have all the optional properties implemented in my non-RDF tool

<niq> is there an earl report of all earl reports that don't contain themselves?

Chaals: Why would anyone do that? If you have an RDF tool it handles all RDF, if not then why would you use an RDF tool to read the schema and generate the table for you, instead of reading the table

SAZ: What about having another appendix with the verbose stuff, and the abbreviated schema being normative

CMN Would get rid of the N3 and the verbose stuff - the smaller the spec, the mnore chance it has of getting read.

<niq> chaals dances on the table

RESOLUTION: Optional properties will be dropped from the shema, and the n3 will be dropped from the spec

<niq> [


Chaals: Some things outstanding - conformance to EARL (tools, doccumments, ...)

SAZ: CarlosI has an action item to write about test subjects
... Nick might be adding some stuff about his thoughts on hash

Conformance likely to be clearer when we ahve the new schema in place.

<niq> too late in the day - my thoughts on hash have gone to pot

<velasco> bentoweb.org

Test Case Description Language (TCDL) by BenToWeb project

<JibberJim> CV: bentoweb EU funded to work on EARL and WAI test suites and improve algorithms for language simplicity colour blindness etc.

<JibberJim> CV: TCDL - test case description language, XML not RDF

<JibberJim> CV: 4 major components in testCaseDescription formalMetadata, technology, testCase and rules

<JibberJim> CV: for every testcase there are successCriterion

<JibberJim> CV: a ruleset contains guidelines and success criteria

<velasco> rulesets URL: http://bentoweb.org/refs/rulesets.xml

<JibberJim> CV: points to specs to identify the successCriterion for the ruleset

<JibberJim> CV: for every test case you have the formalMetadata which is description and Dublin Core data and the source - where the testCase came from

<JibberJim> CV: Then there's the recommendation, which is what mark-up language is the test relevant too

<JibberJim> CV: rules, a collection of rules to meet the test, locations and expected results and the outcome etc.

<JibberJim> CV: test describes the purpose and preconditions of the test and the files to be used, as well as other info required to run the test

<JibberJim> CV: example file will be on the web soon...

<JibberJim> CV: created 400 testcases for WCAG 2

<JibberJim> SAZ: Why not EARL results?

<JibberJim> CV: because Earl wasn't stable

<JibberJim> CV: We're using location based on line/col and xpath

<JibberJim> CV: we opted for xml as we could parse it, which we couldn't RDF

<JibberJim> CV: We could map it to RDF with xslt for public consumption

<JibberJim> SAZ: Will you adopt EARL more actively?

<JibberJim> CV: Absolutely, we were just on a tight deadline and to start quickly.

<JibberJim> CV: the earl:requirement will be the same as our ruleSet

<JibberJim> SAZ: Does anyone want to work on test suite for WCAG?

<JibberJim> chaals: I'm not too excited by test suites, and wcag 2 seems like something santa might bring

<JibberJim> chaals: more exiting would be test suites that are in WCAG 1 which are likely to be in 2

<niq> wcag isa can of rancid worms

<JibberJim> SAZ: we want to review wcag 2 tests...

http://www.w3.org/TR/WCAG20/ is teh current public draft

It's, umm, not very clear.

<JibberJim> SAZ: There's life beyond EARL

<JibberJim> chaals: but we should finish EARL first...

<JibberJim> SAZ: we all agree a Test case description language is important, but we should postpone it until EARL is done.

<JibberJim> chaals: UAAG and ATAG are interesting and valuable future direction

<JibberJim> chaals: Other area of interest is PF but it's all private so we can't talk to them...

<JibberJim> chaals: They've published a roadmap for DHTML Accessibility at http://www.w3.org/WAI/PF/roadmap/DHTMLRoadmap092305.html

<JibberJim> SAZ: time change of meeting...

<JibberJim> SAZ: we should attract more people if possible...

<JibberJim> ACTION: Shadi to investigate possible meeting times [recorded in http://www.w3.org/2005/10/18-er-minutes.html#action03]

<JibberJim> SAZ: Thanks!

<JibberJim> SAZ: Meeting Adjourned.

<shadi> http://www.w3.org/2002/09/TPOverview.html

<niq> saz: next meet cannes can-can ?

<shadi> technical plenary 27 feb - 3 march

<shadi> if not, possibly in spain (invitation by CTIC) some time around that date

Summary of Action Items

[NEW] ACTION: CarlosV get some shape to an EARL guide [recorded in http://www.w3.org/2005/10/18-er-minutes.html#action02]
[NEW] ACTION: jk to update proposal for recording http transaction info [recorded in http://www.w3.org/2005/10/18-er-minutes.html#action01]
[NEW] ACTION: Shadi to investigate possible meeting times [recorded in http://www.w3.org/2005/10/18-er-minutes.html#action03]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2005/11/07 16:21:20 $