IRC log of erswad on 2002-06-25

Timestamps are in UTC.

08:23:39 [RRSAgent]
RRSAgent has joined #erswad
08:23:49 [Zakim]
Zakim has joined #erswad
08:27:07 [wendy]
wendy has joined #erswad
08:28:39 [nadia]
nadia has joined #erswad
08:29:01 [chaalsXXX]
eric? sean?
08:29:26 [libby]
libby has joined #erswad
08:29:30 [Hixie]
Hixie has joined #erswad
08:31:44 [danb_lap]
danb_lap has joined #erswad
08:31:51 [nmg]
nmg has joined #erswad
08:33:12 [wendy]
http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Jun/0038.html
08:33:31 [maxf]
maxf has joined #erswad
08:34:00 [libby]
photos from yesterday: http://swordfish.rdfweb.org/photos/2002/06/24/
08:34:16 [chaalsBRS]
In room, but not on IRC: Dave Pawson, Jim Ley, Nick Kew
08:36:20 [Bert-lap]
Bert-lap has joined #erswad
08:36:58 [libby]
another view from rdfauthor of danbris example: http://swordfish.rdfweb.org/people/libby/rdfweb/earl/danearl.svg
08:37:16 [wendy]
http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Feb/att-0031/01-0.95byExample.html
08:37:54 [chaalsBRS]
http://www.w3.org/2002/servlet_45587-2 two claimpackages (which actually conflict in what they say)
08:42:17 [wendy]
nh object contained all testsubject stuff
08:42:23 [wendy]
dp use language from test area
08:43:30 [libby]
doesn't it work on macos 9?
08:43:58 [wendy]
wac assertedby property on claimpackage.
08:44:58 [danb_lap]
chaals, can you run java?
08:45:06 [libby]
chaals: numerous claims for each tool
08:47:10 [wendy]
wendy has joined #erswad
08:47:31 [wendy]
http://valet.webthing.com/page/
08:49:25 [Hixie]
sample earl data: http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Jun/0040.html
08:53:48 [libby]
wendy: different levels of granularity - could be page level or smaller. need to do both
08:54:09 [libby]
...e.g. a whole website, a page on teh website, or a simgle elemnt on the page
08:58:43 [libby]
chaals: each claim needs to have a confidence rating
08:58:52 [libby]
[discussion of reification]
08:59:24 [libby]
nadia voes we just rename the reified predicates
09:00:58 [libby]
danbri: rdf reification is only good for very simple claims. reifying graphs (say, in the future) would be hard using rdf reification.
09:01:17 [nadia]
from the main earl page:
09:01:20 [nadia]
:Bob earl:asserts
09:01:20 [nadia]
[ rdf:subject :MyPage;
09:01:20 [nadia]
rdf:predicate earl:passes;
09:01:20 [nadia]
rdf:object :WCAG10P1 ];
09:01:26 [nadia]
http://www.w3.org/2001/03/earl/
09:01:55 [nadia]
if you look at the old n3 examples, they really liked the simple [:MyPage earl:passes :WCAG10P1] thing...
09:02:16 [nadia]
which looks neat in n3... but doesn't work as well in practice
09:04:05 [libby]
[we wonder whether to slip inrto groups for a while]
09:04:14 [libby]
slip?
09:04:20 [libby]
er split
09:04:26 [nadia]
pop!
09:05:19 [libby]
danbri/nadia/nmg could go argue about reification and danbri's new proposal
09:05:36 [libby]
wendy/chaals could talk about usecases
09:05:48 [chaalsBRS]
chaals is going to look over the use cases and see what we need EARL to provide.
09:06:38 [chaalsBRS]
(In particular, is there anything that requires it be RDF? Is there any need for the reificiation? What kind of queries do we want to support easily? Can we deal with some of those through an XML representation (isomorphism)?
09:06:42 [chaalsBRS]
etc...
09:07:20 [danb_lap]
oh, my previous attempt at removing reification via using hypertext: http://lists.w3.org/Archives/Public/www-archive/2001Dec/0099.html
09:08:03 [nadia]
rdf is cool because there are growing databases and query systems, and we can contribute to the semantic web!
09:08:22 [danb_lap]
this tried to break out the quoted document into separate docs. I'm thinking the opposite now, sort of...
09:36:59 [chaalsBRS]
A use case: on a vu un contrat?
09:37:06 [chaalsBRS]
(oops)
09:37:19 [chaalsBRS]
Augmenting automatic test results with information from people.
09:37:37 [chaalsBRS]
Allowing a tool to ask for human evaluation of tests, and then moderating a future test overthe same material with the results provided by human evaluation. This uses the fact that EARL assertions are made by a named person/tool, so it is possible to specify whose results (for a given set of tests, or in general) are to be more trusted. HiSoft's Interview wizard, TAW do something like this.
09:37:50 [chaalsBRS]
another use case
09:38:05 [chaalsBRS]
Finding out a test is no good.
09:38:19 [chaalsBRS]
Using a handful of test results to generate a higher level result, we suddenly discover that one of the tests is unreliable. We need to change the test suite, and to be able to identify results based on the bad test and therefore know not to trust those.
09:42:04 [Hixie]
can you use EARL to grade students? :-)
09:45:00 [chaalsBRS]
Sure...
09:47:09 [wendy]
http://www.w3.org/WAI/ER/EARL?
09:48:49 [chaalsBRS]
a [ foaf:mbox <mailto:chaals@w3.org> ] earl:fails deptOfEd:kindergarten
09:49:07 [wendy]
http://www.w3.org/WAI/ER/2002/06/25-earl.html
09:49:55 [Hixie]
chaalsBRS: at the time you create that test result, you're unlikely to have mailto:chaals@w3.org as your URI... ;-)
10:02:40 [maxf]
maxf has joined #erswad
10:10:53 [chaalsBRS]
New Use case:
10:11:08 [chaalsBRS]
Need to meet a new set of requireemnts
10:12:05 [chaalsBRS]
Query existing results, using a new expression of how to derive a result, to see if there is any new testing missing?
10:12:57 [chaalsBRS]
wondering about a use case that requires confidence ratings...
10:13:01 [chaalsBRS]
and how to use them.
10:13:08 [Hixie]
well
10:13:19 [Hixie]
yeah
10:13:28 [Hixie]
the only use case i can think of is covered by Can't Tell
10:13:41 [chaalsBRS]
(So far I have seen them used in Page Valet to indicate how commonly things that might cause problems actually do - which strikes me as belomging to test description rather than results)
10:14:20 [nadia]
nadia has joined #erswad
10:24:05 [chaalsBRS]
Some questions about implementing EARL:
10:24:19 [chaalsBRS]
What should I say about the results of an individual test?
10:24:28 [chaalsBRS]
Can I aggregate tests? How?
10:24:35 [chaalsBRS]
Can I represent results?
10:24:54 [chaalsBRS]
How do I store intermediate results that apply to something going to change?
10:25:21 [wendy]
cmn - i updated earl.html. some new place holders for use cases. i think 12 use cases in all now.
10:25:24 [chaalsBRS]
How do I get existing results and merge them with new tests?
10:25:32 [chaalsBRS]
wendy: cool...
10:25:50 [chaalsBRS]
is thinking about a document that says "how to implement EARL"
10:29:26 [wendy]
ih - why is confidence enumerated? why isn't it numerical?
10:29:52 [Hixie]
...specifically, some hardware testers will have very specific confidence levels
10:30:11 [Hixie]
...e.g., "this test fails (confidence 75.44%)"
10:30:58 [Hixie]
wendy - you can't average these numbers, i'm worried people would do that though
10:37:54 [wendy]
ih perhaps hardcode confidence levels as - 100%, 66%, 33%, 0%?
10:38:32 [Hixie]
ih - so you could use numbers if you know them, but if you just want a vague idea (like in page valet) then you would use the constants
10:39:34 [Hixie]
ih - if you have numeric confidence levels, you can remove the Can't Tell result (it's equivalent to one of the other results with confidence: 0%)
10:40:54 [Zakim]
Zakim has left #erswad
10:54:45 [nadia]
nadia has joined #erswad
11:07:22 [wendy]
lunch!!!!
11:07:24 [wendy]
eat you foolls!
11:07:32 [wendy]
yum yum yum
11:10:55 [maxf]
maxf has joined #erswad
12:04:20 [maxf]
maxf has joined #erswad
12:12:53 [wendy]
can we get rid of daml+oil by using XMLSchema instead of a DTD to further constraints document instances? question is: is it an aggregation issue or a document issue (ala Libby)?
12:13:52 [wendy]
hello? are you all ready to start or do you want another 15 minutes?
12:13:58 [libby]
libby has joined #erswad
12:14:00 [danb_lap]
don't need to get 'rid of' d+oil. They do different jobs... XML Schema talks about types of document, DAML+OIL talks about the things described in those documents.
12:14:37 [libby]
15 what?
12:14:48 [nadia]
m&ms
12:14:53 [libby]
ah
12:20:15 [nmg]
EARL examples from schema breakout - http://www.ecs.soton.ac.uk/~nmg/er-swad/
12:22:27 [Hixie]
nmg: hey, i didn't assert that! :-P
12:22:44 [Hixie]
nmg: do you have a reference for the genid: scheme?
12:24:38 [maxf]
http://lapin-bleu.net/cmnmnm.jpg
12:24:39 [danb_lap]
thanks nmg
12:25:09 [danb_lap]
I think we need a way of constraining EARL extensions / descriptive practcie so that the subject node doesnt get smushed
12:29:47 [libby]
cool!
12:29:55 [libby]
not the sitting back, particularly
12:30:12 [libby]
I've done some rdf pictures of nmg's examples: http://swordfish.rdfweb.org/people/libby/rdfweb/earl/
12:30:46 [danb_lap]
nmg, do you ave an example where the testsubject is given an actual uri?
12:42:42 [nmg]
http://www.ecs.soton.ac.uk/~nmg/er-swad/earl-example1.rdf
12:44:45 [nmg]
(to danbri: http://www.ecs.soton.ac.uk/~nmg/er-swad/earl-frag2.xml)
12:45:18 [wendy]
WAKE UP!!!!
12:48:30 [chaalsBRS]
Libby's summmary of schema discussion:
12:48:48 [chaalsBRS]
We don't like the subject/predicate/object properties used to reify a little bit of the graph.
12:49:11 [danb_lap]
danb_lap has joined #erswad
12:49:18 [chaalsBRS]
It's ugly, and potentially not sufficient, and there are no tools that rely on using the RDF property names.
12:49:43 [chaalsBRS]
It isn't that we are dumping the idea, just that we use more appropriate properties.
12:50:14 [chaalsBRS]
subject is what we are testing, object is the testcase, and predicate is the result.
12:50:31 [chaalsBRS]
So we propose to use something like that...
12:51:11 [chaalsBRS]
--
12:51:21 [nmg]
slightly different diagram of what libby's talking about: http://www.ecs.soton.ac.uk/~nmg/er-swad/earl-strawman.gif
12:51:25 [chaalsBRS]
Danbri has a testcase that shows problems which occur with this amount of reification.
12:51:34 [nadia]
predicate is being replaced with "result" or something similar
12:51:50 [nadia]
instead of being a subclass of rdf:Predicate, it is now just a normal property
12:51:56 [nadia]
that can have other properties more easily assigned to it
12:52:04 [nadia]
and won't kill the existing rdf engines
12:52:29 [chaalsBRS]
NK. I'm considering using RDF to define tests. How does this affect doing that?
12:52:41 [chaalsBRS]
LM This is just changing names mostly.
12:52:50 [nmg]
earl:result is an instance of rdf:Property, and has a range of earl:TestResult
12:53:23 [nmg]
previously, rdf:predicate was locally constrained to have a range of earl:ResultProperty, which was a subclass of rdf:Property
12:53:27 [chaalsBRS]
NH This doesn't really affect that, although there are some useful properties within EARL that you might re-use
12:54:50 [chaalsBRS]
CMN changing the names allows us to explain what the properties are actually for.
12:55:07 [chaalsBRS]
...also, it would be interesting to apply XAG to EARL (apropos of nothing...)
12:55:19 [chaalsBRS]
LM: Next thing is more controversial...
12:55:40 [chaalsBRS]
... making the thing being tested be identified as a bNode
12:56:05 [chaalsBRS]
DB EARL tries to categorise things into controversial claims and uncontroversial statements.
12:56:30 [chaalsBRS]
... controversial are things like "this tool doesn't work" - it carries information about who carried the claims
12:56:52 [chaalsBRS]
uncontroversial stuff is things like some page exists.
12:57:49 [chaalsBRS]
...problem is when it is possible to merge uncontroversial things in troublesome ways. For example a page is tested. In one test it has a size of 444 bytes, and in another test it has 555 bytes but in both cases it has the same URI.
12:58:42 [nmg]
http://www.ecs.soton.ac.uk/~nmg/er-swad/
13:00:36 [libby]
svg version: http://swordfish.rdfweb.org/people/libby/rdfweb/earl/earl-frag2.svg
13:00:54 [chaalsBRS]
So it merges the thing saying that there is a page which is both 444 bytes and 555 bytes.
13:03:59 [libby]
can preserve this info by mading people point to 'bnodes', i.e. point indiectky at the uri, using an intermediate node
13:04:08 [libby]
(see frag3)
13:04:13 [libby]
...*but*
13:04:30 [libby]
- a bit of a mixed message (we often say, use uris)
13:05:01 [libby]
- plus, daml:unambigous propery (also in webont) which danbri and libby are often using, and which is v useful - can merge these 2 nodes right back again!
13:05:36 [libby]
chaals: we will definely want to talk about webpages and those webpages will change. should we being doing this - or webont...?
13:06:02 [libby]
danbri: we should defubuelty get backj to webont, but nt their fault
13:06:20 [libby]
...it's a touch problem, however we model it (rdf or not)
13:06:26 [libby]
s/touch/tough
13:07:05 [libby]
chaals: the case occurs where you have an unambigousproperty on the node - can you still merge?
13:07:15 [libby]
danbri: they can;t merge, I think
13:08:09 [libby]
chaals - does unambig prop w diuffeent values drop the merge?
13:09:11 [libby]
nmg: if 2 objects have same val for a prop that's unambigius, they *must* be the same object
13:09:27 [libby]
...uniqueproperty must only ahve one value (that's different)
13:10:09 [libby]
danbri: in the example case, we are not saying tehre is unambigprop - reprof is not unambiguous, so the problem doesn;t occur
13:10:27 [libby]
chaals: so we have to not say that they're unambigous
13:11:07 [libby]
danbri: maybe 'versionOf' a software tool, and the softwaretool could have an 9unabmigusprop) homepage
13:12:35 [libby]
jim: we've disxcusse all this about 6 months ago
13:13:19 [libby]
...we are actually already doing this...
13:13:33 [libby]
...addressed alot of these point before with sbp
13:14:10 [libby]
...alreday do something like reprof
13:15:35 [libby]
...is this ever going to occur?
13:15:55 [libby]
danbri: maybe...extending without talking to us, e.g. if people plugged in unambiguous property
13:16:23 [libby]
...danbri is writing this doen
13:16:51 [libby]
chaals: wants to do a document on how to implement earl
13:17:09 [libby]
...people find it hard
13:17:21 [libby]
..useful thing to do
13:17:57 [libby]
...e.g. what kind of propettyies shoudl I be useing for a test; what sort of stogae shoudl use, how shluld I extend earl etc
13:18:25 [chaalsBRS]
Libby continues with the report.
13:18:36 [wendy]
(refer to cmn's notes starting around 06:21)
13:18:54 [libby]
...need to talk to sbp about this next:
13:19:05 [chaalsBRS]
NG: Assertion used RDF reification mechanism and vocabulary.
13:19:22 [chaalsBRS]
... side effect was that the result type had to be of type property
13:19:43 [chaalsBRS]
... so the result property was a subclass of type property which is problematic for lots of tools.
13:19:43 [libby]
nmg: with rdf reified syntax, the type of the predicate is type property, so the result proprty had to be a subclass of rdf property - somewhat problematic as nmg mentionned yesterday
13:19:57 [chaalsBRS]
... removing reified syntax means that we can solve that.
13:20:14 [chaalsBRS]
... You have a result property (like passes) that has properties itself.
13:20:25 [libby]
..we can remove dependance on rdf:property (and on daml, incidently)
13:20:38 [nmg]
http://www.ecs.soton.ac.uk/~nmg/er-swad/earl-strawman.gif
13:20:57 [libby]
...we get an object of type testresult rather than a property
13:21:18 [nmg]
http://www.ecs.soton.ac.uk/~nmg/er-swad/earl-example1.txt
13:22:24 [libby]
...proposigonm one and only one confidence and one and only one validity
13:22:35 [libby]
...not a single-instance object as before
13:22:43 [libby]
...simplifies a bit; extensible
13:23:00 [libby]
...cleaning up after removing reficiation
13:23:37 [libby]
wendy: what about adding 'severity' - type of failure
13:24:09 [libby]
...so if we know if fails, then we can do easy comparison
13:24:23 [libby]
nmg: can still ask that, but sllightly different query
13:24:51 [libby]
...makes it easy to add new explantory results, e.g. a message field
13:25:04 [libby]
wendy: thinks might solve an extension issue they've had
13:26:42 [wendy]
http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Apr/att-0039/01-sample.4.rdf
13:27:15 [libby]
mng: yes: we made it a proprty of the result
13:27:23 [libby]
chaals: with merging?
13:27:43 [libby]
nmg: yes, ok, unlike sbp's versioon, the result objects are different objects
13:27:48 [libby]
..so can merge ok
13:28:33 [libby]
wendy: there is a problem with the merge and pass/fail?
13:29:20 [libby]
nmg: before: was an instance of a certain property, so couldn''t add a meaage to it, because would be a property of all the properties
13:30:03 [libby]
mng: the 'pass' should be an instance of the validity level - an object
13:31:08 [libby]
[nmg draws stuff on flipchart]
13:32:33 [libby]
basically, the resultproperty changes from a property to an object - a TestResult
13:32:37 [libby]
...gets rid of some duplication
13:32:49 [libby]
...still ahve predeifined properties for pass and fail
13:33:15 [libby]
- not very controversial
13:34:12 [libby]
...range of predicate (now 'results') is no longera property
13:34:22 [libby]
...could still write a rule
13:39:01 [nadia]
libby: suggests constraining earl syntax so we can be like rss and write a dtd and an rdf schema
13:39:05 [chaalsBRS]
q+ niq
13:39:31 [nadia]
then in situations where we have confidence in the assertions we'll be dealing with, we'll be able to use the simpler syntax
13:39:41 [Zakim]
Zakim has joined #erswad
13:39:59 [nadia]
or in more rdf-type situations where we need to deal with judging the validity of our assertions, we deal with the more complex rdf-type earl
13:40:34 [libby]
nk: valet actually does this - has an internal xml format -> RDF when shared
13:41:38 [libby]
dp: what's the point of haviong 2 dforms if you lose ifnormation?
13:41:55 [libby]
danbri: simple form easier to read and index; verbose format very very careful
13:43:11 [libby]
dp: so you can;t go from simple xml-> complex xml (rdf) because missing information - you could do this if you passed it a parameter
13:43:35 [libby]
wendy: also - if constrained xml then that allows to go back and forth
13:44:26 [libby]
dp: has a dtd
13:46:20 [libby]
[some agreemne that both 1) the constrained RDf syntax (RDF+DTD) might be a good idea, and that 2) an XSLT transform between a simple and complex version might be good ideas]
13:46:39 [danb_lap]
http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Jun/0041.html -- my rough vocab design issues writeup
13:46:49 [libby]
chaals/dp: would be good to have a common definition of property names etc
13:49:59 [wendy]
http://faff.com/
13:51:50 [nadia]
http://lists.w3.org/Archives/Public/www-archive/2002Jun/att-0035/01-simpleEarl.xml
13:51:53 [nadia]
i sent the mail, but it hasn't appeared yet
13:54:10 [libby]
(this is dp's dtd)
13:55:35 [libby]
wendy: is there consensus on all this proposed (among the meeting)?
13:56:28 [libby]
wendy: it sounds like a dtd would be a good thing - consensus.
13:56:32 [libby]
...restaing the question:
13:56:52 [libby]
1) gettign rid of rdf reification names
13:57:03 [libby]
2) using bnodes for the subject
13:57:14 [libby]
3) nmg's change for results
13:57:33 [libby]
4) using a constrained RDF syntax (with a DTD)
13:57:57 [libby]
5) having a simple and a complex version, mappable with XSLT
13:58:13 [libby]
wendy: consensus on 1,2,3?
13:58:33 [libby]
...check with sbp first; then modify the schema
13:59:23 [libby]
[we look at sbp's summary to wendy about the schema]
13:59:56 [libby]
ACTION wendy to talk to sbp about these changes
14:00:57 [libby]
wendy: someone from teh RDF interest group needs to take this merging problem to the RDFIG
14:02:34 [libby]
...when we get back, generate earl for all the new user scenarios we came up with
14:03:07 [libby]
--coffee--
14:31:26 [nmg]
http://www.ecs.soton.ac.uk/~nmg/er-swad/earl-example1.txt
14:35:16 [nadia]
http://lists.w3.org/Archives/Public/www-archive/2002Jun/att-0037/01-simpleEarl.xml
14:35:16 [nadia]
there
14:45:40 [wendy]
in dave's dtd: replace testcriteria with testsubject
14:49:14 [wendy]
going through dave's dtd: http://lists.w3.org/Archives/Public/www-archive/2002Jun/att-0035/01-simpleEarl.xml
14:49:21 [wendy]
discussing grouping tests, suites
14:49:35 [wendy]
nk suite should be property of the assertor, as it applies it.
14:49:46 [wendy]
cmn separate suite out completely
14:49:52 [wendy]
jl exactly.
14:50:55 [wendy]
cmn suite should have property 'includes" some test case, rather than testcase saying i believe to some suite.
14:52:07 [wendy]
dp i was grouping tests. what i'm testing against doesn't need referencing.
14:52:41 [wendy]
cmn either search every testcase (to ask if it belongs to wcag). want something that says "these things belong to wcag"
14:52:53 [wendy]
jl even then don't know if you haven't done them.
14:53:12 [wendy]
dp that seems to be a configuration file.
14:53:26 [wendy]
cmn yes. that separate config is useful to have in an earl report, but not at the level of testcase.
14:53:36 [wendy]
dp yes, and support what i'm testing with t his test.
14:53:50 [wendy]
cmn test does not know if it is attached to a suite or not.
14:53:55 [wendy]
dp for dtbook it is useful.
14:53:59 [wendy]
..for traceability.
14:55:04 [wendy]
cmn that is in a test description language.
14:55:17 [wendy]
.. if you query that test then you can determine...at the EARL level don't need.
14:55:38 [wendy]
.. your use case is in a system where tests describe what they test.
14:55:56 [wendy]
.. what are they testing, how test being run, what suite part of, etc. earl just gives you the result of the test.
14:56:05 [wendy]
jl you have the test id. then externally, you can get the rest of the info.
14:56:25 [wendy]
dp execution info?
14:56:39 [wendy]
cmn i argued that there are 2 useful things re: result:
14:56:46 [wendy]
1. who produced it. (trust negotiation)
14:57:25 [wendy]
2. is that result an atomic result or derived from aggragating results of tests (e.g. all tests for checkpoint 1, therefore pass cp1)
14:57:35 [marja]
marja has joined #erswad
14:57:39 [wendy]
cmn the particularly info that allows you to do aggregation should be externalized.
14:57:56 [wendy]
cmn claim, assertor - at that level test suite consists of...
14:58:03 [wendy]
dp test environment?
14:58:08 [wendy]
cmn at level of assertor.
15:00:45 [wendy]
nk when i've implemented tests, i've reimplemented for each suite. i have a pointer back to the href in wcag or 508 as a property of the test.
15:01:35 [Hixie]
(the W3C selectors test suite considers of many small test suites which use different sets of the same tests)
15:01:39 [Hixie]
(fwiw)
15:02:35 [wendy]
nk i could have it separate.
15:02:47 [wendy]
exclusion?
15:02:55 [wendy]
cmn this is part of test description.
15:03:15 [wendy]
cmn we ran this test, we excluded x from the test.
15:04:39 [wendy]
dp have that info in the test case...what is that file called? test config file?
15:05:07 [wendy]
cmn suite: this suite contains these tests.
15:05:48 [wendy]
proposal: move test description info to separate file.
15:05:58 [wendy]
ng if still have suite, still have level?
15:06:05 [wendy]
jl perhaps in "i have collection of tests"
15:06:13 [wendy]
dp this test is really important (in config file)
15:06:38 [wendy]
cmn AA contains this set of tests+ level A. then level A is this set.
15:07:10 [wendy]
cmn if compare 508 and wcag. 508 doesn't have idea of level. in wcag there are tests are pri 2 that are in 508. thus diff levels in diff requirements, but exactly the same test.
15:07:59 [wendy]
ng if excludes and level are no longer in earl, what is testproperty.
15:08:26 [wendy]
s/testproperty/left in testcriteria?
15:08:29 [wendy]
id
15:08:42 [wendy]
dp how do you tell if the test has passed or failed
15:09:50 [wendy]
proposal: remove testCriteria (move into test spec domain - outside of earl)
15:09:52 [wendy]
consensus in the room.
15:10:34 [wendy]
operator instructions?
15:10:42 [wendy]
jl can still say if have done automatically or not.
15:10:50 [wendy]
dp measure in pixels, width of x.
15:11:03 [wendy]
ng but this is language to report the result.
15:14:16 [wendy]
dp don't have measured value, which could be user input.
15:14:19 [wendy]
nk we have earl:comment
15:14:37 [wendy]
wac not convined to get rid of operator instruction. might not be the best place.
15:15:12 [wendy]
cmn in the note, (testing if image fits on screen) "it fits, but only just."
15:15:38 [wendy]
cmn if rely on that for doing further extensions, at this level, we are asking for trouble.
15:15:49 [wendy]
dp i'd like to isolate actual measured value and note.
15:15:57 [wendy]
jl you can do that in your private namespace if it applies to you.
15:17:07 [xower]
xower has joined #erswad
15:17:29 [wendy]
dp instructions to operator, and response from operator.
15:17:43 [wendy]
ng which is why can add properties to add to test result.
15:17:58 [wendy]
cmn comments could be rdf and all sorts.
15:18:07 [Hixie]
(btw, wendy, when you have a sec, see /msg)
15:18:18 [wendy]
yep. can i display onscreen? :)
15:18:31 [Hixie]
sure :-)
15:19:23 [wendy]
proposal: get rid of excludes, and level
15:21:04 [wendy]
@@add note as a property of TestResult (if not there already...can't seem to find)
15:21:23 [wendy]
proposal: delete purpose
15:24:28 [wendy]
failurePoint
15:24:36 [wendy]
cmn that's where fuzzy pointer to failure goes.
15:25:31 [wendy]
cmn you have to identify where the problem arose
15:25:47 [wendy]
cmn "inside this page at this point is where it failed"
15:26:10 [wendy]
jl you have a whole range of tests to arrive at the conclusion, you have a range of parts of the page that you test.
15:26:26 [wendy]
.. all of these tests subjects ...
15:26:41 [wendy]
nk look at page valet results, they are point in the structure not the whole page.
15:26:55 [wendy]
cmn works if you can split a page node by node.
15:27:13 [wendy]
cmn not all tests at children of a page. similarly, running a tool.
15:28:33 [wendy]
jl if you invalidate that one part, you have to rerun entire suite.
15:28:51 [wendy]
cmn if you have location info, can fix at that location, know it has no other impacts.
15:30:28 [wendy]
nh since it's 2 bits of info, why not put the relevant in each. if any more info that you need, depends on your use. it could be note.
15:30:42 [wendy]
dp e.g. - dtbook. 500 SMIL files. need file, location to say what's wrong.
15:30:50 [wendy]
jl that's the test subject. that test subject failed.
15:31:07 [wendy]
cmn you have 500 results. 499 pass, 1 fail.
15:31:21 [wendy]
.. requires that you know that dtbook contains 500 sep objects.
15:32:24 [wendy]
Action: DP, CMN, JL discuss where this should go.
15:33:26 [chaalsBRS]
(where this == location of a problem identified, or the thing that causes a page to pass)
15:34:02 [wendy]
wac add severity?
15:34:24 [wendy]
ih importance is property of test, result can have certain amount of success. e.g. crash is a failure w/high severity.
15:34:32 [wendy]
dp operator note?
15:35:01 [wendy]
ih function of test, it determines how to do it. could be automated...like confidence. when you find pass/fail, you also note hoe confident.
15:35:05 [wendy]
nk already defined.
15:35:31 [wendy]
action ih: send something to the list re: severity (define it)
15:39:00 [wendy]
do we need operator?
15:39:07 [wendy]
cmn if you want to ring up the person who ran the tool.
15:39:33 [wendy]
wac if the assertor is a tool, operator is the person who ran the tool.
15:40:29 [wendy]
action wac define operator ala prev discussion.
15:40:36 [Zakim]
Zakim has left #erswad
15:40:53 [wendy]
make clear that operator is a property of assertor.
15:41:05 [wendy]
(in the doc, it's clear in the schema)
15:41:34 [wendy]
action wac: email dtd (with changes) to the list
15:44:36 [wendy]
criticalness of test is in test description not in earl (delete @@schema from user scen)
15:46:52 [wendy]
test enviroment - in dp's dtd is "testenv" at assertor level.
15:48:23 [wendy]
only 2 remining @@schemas on user scenarios.
15:52:18 [wendy]
going through sbp's issues.
15:52:19 [wendy]
http://www.ietf.org/rfc/rfc2616.txt
15:52:37 [wendy]
(have covered most of them today. but issue about "document source" in uaag ...
15:53:58 [wendy]
what's the diff between a note and a rec?
15:54:05 [wendy]
danbri "about 2 years and a nervous breakdown"
15:54:37 [wendy]
cmn value of rec is higher.
15:54:40 [sbp]
heh, heh
15:56:09 [wendy]
it is included in charter: Coding may include writing a schema to be used by a variety of tools to store evaluation and repair data about a page or site.
15:56:38 [wendy]
but doesn't say anything about documenting it. :)
15:56:52 [sbp]
coding includes documentation! :-)
15:57:16 [wendy]
:)
16:02:45 [wendy]
cmn maybe take back to w3c to find out - is there interest (in going to rec)? should we talk w/qa - do they have the resources?
16:02:51 [wendy]
may publish as note and drop.
16:04:35 [wendy]
-- splitting into 2?
16:04:48 [danb_lap]
(or publish as note and take stock at that point?)
16:04:49 [sbp]
2 what?
16:04:51 [wendy]
nh renaming makes structure more clear. less of an issue.
16:04:58 [wendy]
into 2 schemas.
16:05:02 [sbp]
oh
16:06:18 [nadia]
taking test definition properties out of the schema makes it cleaner too
16:06:21 [wendy]
we oota here
16:06:51 [danb_lap]
------
16:07:00 [danb_lap]
RRSAgent, pointer?
16:07:00 [RRSAgent]
See http://www.w3.org/2002/06/25-erswad-irc#T16-07-00
16:07:07 [nmg]
nmg has left #erswad
16:07:11 [sbp]
403
16:07:40 [danb_lap]
done.
16:07:41 [danb_lap]
ok?
16:08:49 [sbp]
done. cheers
16:09:30 [wendy]
see ya sbp.
16:10:22 [chaalsBRS]
seeya folks