11:09:22 <RRSAgent> RRSAgent has joined #er
11:09:33 <olivier> good bot
11:09:52 <chaals> So, making MUTAT work nicely...
11:10:32 <olivier> got the source?
11:10:48 <chaals> http://www.w3.org/QA/Tools/MUTAT/
11:10:59 <chaals> (not the source but the tool...)
11:11:20 <olivier> http://dev.w3.org/cvsweb/perl/modules/W3C/MUTAT
11:11:23 <olivier> the source
11:11:53 <olivier> I read the source on friday, it's OK (for my standard at least)
11:12:22 <olivier> I noticed a few things, but nothing really bad
11:12:39 <olivier> need to change a few things that are still linked to the service @barwired
11:13:18 <chaals> When you enter the WCAG checker page (as an example) http://www.w3.org/QA/Tools/MUTAT/bin/mutat?script_status=1&TEST_url=http://www.w3.org/QA/Tools/MUTAT/wcag.n3
11:13:33 <chaals> it doesn't provide an option to say what page you are testing...
11:15:09 <olivier> it's not necessarily a page, right? But I understand what you mean, you can give properties of the test subject but not its URI 
11:15:32 <chaals> Right. So I would like to make it do that.
11:15:55 <chaals> I'd also like to make it post results as annotations to http://earl.w3.org
11:16:24 <chaals> (At the moment it generates its own URI in the mutat namespace for the testSubject)
11:16:39 <olivier> hmm re annotations, wait a minute, there's something in the code
11:16:42 <chaals> s/testSubject/regarding/
11:17:08 <olivier>  my $request = HTTP::Request->new(POST, 'http://iggy.w3.org/annotations', $h, &create_output);
11:17:11 <chaals> Yep. sub_post_to_database
11:17:19 <olivier> right
11:17:42 <JibberJim> with old EARL (so needs updating), http://jibbering.com/2002/9/earl-assertions.html has bookmarklets to post simple human evaludations to the earl db.
11:18:36 <JibberJim> Win-32 IE only bookmarklets, security kills anything else.
11:18:38 <chaals> At the moment the earlDB uses its own namespace http://earl.w3.org/temporary#
11:19:00 <chaals> Oh. So not much help me having a look now ;-)
11:19:17 <olivier> you machead you
11:19:52 <JibberJim> no, not really, but once we're ready to put evaluations in to build a good test-corpus it's a quick way to get simple feedback from people I think, I just press a link on my toolbar and the db has an evaluation.
11:20:52 <chaals> jibberjim: so you're updating it now to use temporary earl, right? ;-)
11:21:27 <olivier> chaals : re annotation, I don't think there's anything to "launch" it. we could make it an option to the submit (what a bad name) button at the bottom of the test page
11:21:36 <JibberJim> Ah, although I could actually do it like I do the validator.w3.org bookmarklet thingy, which as long as I can get a webpage onto earl, would give us Mozilla at least.
11:21:55 <olivier> i.e [get EARL code] [Send Annotation]
11:22:01 <chaals> jibberjim: and iCab
11:22:10 <chaals> olivier: right
11:22:12 <JibberJim> no, icab doesn't let me...
11:22:25 <JibberJim> I've never found an iCab person to lobby...
11:22:28 <chaals> Oh?
11:22:49 <chaals> olivier: It would be useful if we could query the database for existing EARL, too.
11:23:06 <JibberJim> Only mozilla and IE provide javascript with a way to make HTTP requests.
11:26:37 <olivier> the n3 file?
11:26:51 <olivier> try http://www.w3.org/QA/Tools/MUTAT/bin/mutat
11:26:55 <olivier> (no parameter)
11:29:13 <JibberJim> In MUTAT, where does the earl:result come from? http://www.w3.org/QA/Tools/MUTAT/wcag.n3#C
11:32:35 <chaals> OK, found the TEST_url
11:33:14 <chaals> So the bit that does question and answer needs to ask for that.
11:33:35 <olivier> test_url?
11:33:52 <olivier> but I believe you can't build this page without knowing it already
11:34:19 <olivier> and this is why it's asked for beforehand
11:34:38 <chaals> Right. So the n3 test scheme is passing it as a fixed value, rather than asking for it.
11:34:59 <chaals> This means that it is not really useful for testing anything...
11:35:35 <olivier> no, it's not a fixed value
11:35:37 <chaals> (unless you're going to write some RDF somewhere that says whatever the generated value MUTAT gave is a representation of something you wanted to test. At which point you could write the EARL, too...)
11:35:52 <olivier> hmm
11:35:57 <chaals> Well,  it is from a user's perspective.
11:36:05 <olivier> there's a script for this
11:36:12 <olivier> have a look at create
11:36:17 <chaals> If you go to the bare interface you get to input the testSubject.
11:36:42 <olivier> http://www.w3.org/QA/Tools/MUTAT/bin/create
11:36:48 <chaals> (but it doesn't handle XML apparently - it choked when I gave it http://www.w3.org/2001/sw/Europe/
11:37:54 <olivier> aaaaah, I see
11:37:58 <chaals> Ergh. Software error...
11:38:09 <olivier> sorry confused between testuri and test-subject
11:38:18 <olivier> soft error, where?
11:39:08 <chaals> go to http://www.w3.org/QA/Tools/MUTAT/bin/mutat and enter a URI of a Web page
11:39:45 <olivier> it's supposed to be an n3 file, not any page
11:39:53 <chaals> If you are using a test scheme that is already written (which makes sense) you should still be able to say what to run the tests on.
11:40:11 <olivier> yes, but that's not where you put it
11:40:22 <olivier> this input box is for the test scheme
11:40:24 <chaals> Oh, OK.
11:40:44 <chaals> ACTION: CMN make http://www.w3.org/QA/Tools/MUTAT/bin/mutat a bit clearer about what it wants.
11:41:00 <olivier> agreed, rewordings needed
11:41:36 <olivier> BTW on a general basis when you change the source and want it applied on the service you need me (or any sysperson)...
11:42:14 <olivier> unless you have an account on the cgi box
11:42:21 <olivier> which I doubt :)
11:42:29 <chaals> I don't. That's why I wanted you here ;-)
11:42:36 <olivier> okay :)
11:42:54 <chaals> Even if I give it the name of a test file, it doesn't allow me to tell it what I am running the tests against.
11:43:27 <olivier> chaals : right, I think it comes up with a random URI to define this uri
11:43:38 <olivier> instead of asking the user to input it
11:43:45 <chaals> Yep. That's not helpful....
11:44:35 <chaals> ACTION: CMN make it ask the user for the URI to be tested
11:45:06 <olivier> action 2 means changing code in sub user_questions, FWIW
11:45:09 <chaals> This is just getting http://www.w3.org/QA/Tools/MUTAT/bin/mutat to include an extra field...
11:45:36 <olivier> not only this
11:45:49 <chaals> oh, and making the thing use that...
11:45:53 <olivier> yes
11:46:18 <olivier> another thing I'm pondering is GET/POST
11:46:43 <olivier> I discussed this a bit w/ Nadia but she wasn't convinced
11:47:51 <olivier> POST is a pain, especially when debugging...
11:48:46 <danbri> danbri has joined #er
11:48:56 <olivier> hey danbri
11:50:15 <chaals> ACTION: CMN make it use the earl database.
11:52:44 <chaals> Aah. TEST_url isn't the thing I am trying to deal with - that's the URI for the N3 file that describes the test.
11:52:44 <olivier> mine is a piece of paper 10 cm away :)
11:52:51 <olivier> yes
11:53:14 <olivier> (I guess the one confused wasn't me, then)
11:54:48 <olivier> danbri : want an AI or two?
11:55:34 <olivier> like, care to get me an orange juice? :)
11:55:41 <danbri> sure. I'll drink it too.
11:55:45 <olivier> hehe
11:56:03 <danbri> Re EARL, if people bung me some nice .tar.gz of EARL data, I'll do my best to make rdf query demos around it...
11:56:22 <olivier> chaals : I believe there is no variable for the test "page" (not the scheme)
11:56:39 <olivier> the earl lib seems to come up with a random one
11:56:41 <chaals> danbri, if you use earlget.rb you can collect some...
11:57:05 <chaals> olivier: OK, so I don't feel so stupid feeling like I am searching in vain for where it decides that.
11:57:35 <olivier> we can *both* be stupid, at worst
11:58:06 <danbri> coolness, will do.
11:59:26 <chaals> Danbri, there are only a few things there at the moment on http://www.w3.org/ and http://jibbering.com/
11:59:37 <chaals> and a few on some random mutat namespace that got created.
11:59:59 <chaals> But JibberJim might be making a tool that will let you add lots of things pretty quickly
12:03:38 <olivier> {xx} ?
12:03:41 <chaals> Hmm. so my $testsubject_uri = $test->{test}{ns}{'td'} . 'TestSubject' . time;
12:03:48 <chaals> means...
12:03:58 <olivier> embedded hashes
12:04:39 <chaals> in idiot's terms?
12:04:43 <olivier> hmm
12:04:49 <chaals> there is a thing called $test
12:04:55 <olivier> dictionnary
12:05:04 <olivier> you have a term and something this term refers too
12:05:09 <chaals> indide it, there are name/value pairs
12:05:15 <olivier> what it refers too can be anything
12:05:22 <danbri> hashes inside hashes
12:05:27 <olivier> and this anything can be another dictionnary
12:06:06 <chaals> the bit I am wondering is about $test->{test}{ns}{'td'}
12:06:06 <olivier> here you have a hash inside a hash inside a hash
12:06:31 <chaals> So that means $test is a hash, and test is a member which is a hash.
12:06:45 <olivier> $test is a perl object
12:06:46 <chaals> that hash has a member called ns which is a hash.
12:06:53 <olivier> which is more or less a hash
12:07:03 <chaals> and we want the thing in ns whose key is 'td'
12:07:11 <olivier> one of its variable, $test->{test} is a hash, etc
12:07:16 <olivier> exactly
12:07:20 <olivier> you got it right
12:07:23 <chaals> OK.
12:07:52 <chaals> Thanks for the little PERL tutorial.
12:08:07 <olivier> heh
12:09:07 <olivier> perl is easy to write, hard to read
12:09:12 <olivier> ...
12:11:10 <chaals> so it seems to me that the testsubject is invented on the spot in sub create_output
12:11:35 <chaals> (this relates to how to do action 2...
12:11:41 <chaals> ...without breaking stuff)
12:11:54 <olivier> let me check
12:12:41 <chaals> it's a string: namespace, "TestSubject" , time
12:12:47 <olivier> ugh
12:12:53 <chaals> and then it gets used there for the output.
12:12:54 <olivier> got it
12:13:05 <olivier> so the unique ID is 'time'
12:13:11 <olivier> well...
12:13:24 <chaals> Which means it shouldn't be too hard to let a user decide what they are testing.
12:14:13 <olivier> no, it should be fairly easy, you need to amend the form, then do a test for param('testsubject') and if it doesn't exist fallback to the timestamp method
12:14:59 <olivier> I can take the action 2, if you want
12:15:14 <olivier> I now have a fairly good idea of what to do, and I know perl :)
12:16:08 <olivier> ok, fine with me too
12:18:12 <chaals> Hmmm. When it does algae queries it should GET, no?
12:18:37 <olivier> dunno algae, but Nadia seems to dislike GET
12:18:49 <JibberJim> Annotea accepts GET/POST interchangeably.
12:19:17 <chaals> Right. /me uses GET to query algae in http://www.w3.org/2001/sw/Europe/200209/annodemo/earlget.rb
12:19:18 <JibberJim> GET has a 600
12:19:37 <chaals> This is a query that doesn't change anything, so shouldn't be a POST
12:19:43 <chaals> 600?
12:19:43 <JibberJim> 600 character limit for directly entered urls in IE6 so urls with algae in would not be sharable.
12:20:17 <JibberJim> only for directly entered ones, it can deal with them in forms.
12:20:46 <chaals> Well, where this thing is generating its own query string they are less than 600 chars. And we coud always have an errorCheck if it is going  to generate things that are longer...
12:21:10 <chaals> So if the link were in a page IE would be fine if it were 1k long?
12:22:12 <chaals> i.e. <a href="query1800n">You don't want to see this URI</a>
12:22:40 <chaals> Hmm. iCab only allows bookmarks around 256 characters or something annoying, IIRC
12:23:15 <chaals> s/1800n/[[some string of length 1800]]/
12:23:17 <JibberJim> I think so, it's fine if it's generated by a form, not tested <a href="..."> to say, but I believe so.
12:24:25 <chaals> failing which I'll figure out how to proxy useful queries...
12:27:48 <chaals> Anyway, I'll see if it matters.
12:28:08 <chaals> More interswting development would be to allow for updating a particular statement.
12:28:19 <chaals> Which according to the protocol SHOULD be done with a PUT
12:30:02 <chaals> if we complete the current lot of actions I think we will have done well.
12:30:22 <chaals> Enough for everyone?
12:32:38 <olivier> no, no beep :)
12:32:55 <olivier> but I'm here
12:33:33 <olivier> fine with me, esp. since I don7 have any AI
12:33:50 <olivier> just have to be around to review/install
12:34:38 <chaals> ACTION: OT teach chaals PERL
12:35:00 <olivier> uh-oh
12:35:04 <chaals> ACTION: JibberJim make his bookmarklet post to http://earl.w3org/earl
12:35:31 <JibberJim> How do I tell the RRSAgent I've completed my AI?
12:37:10 <chaals> wait until I have implemented a bot that does it.
12:37:50 <chaals> Alternatively, they are avaialble as RDF. You could annotate it with a 'completed' property - I assume there is one in the vocab that RRSAgent uses...
12:38:56 <chaals> Action items: http://www.w3.org/2002/09/30-er-actions.rdf
12:39:11 <chaals> RRSAgent, bye
12:39:11 <RRSAgent> I see 5 open action items:
12:39:11 <RRSAgent> ACTION: CMN make http://www.w3.org/QA/Tools/MUTAT/bin/mutat a bit clearer about what it wants. [1]
12:39:11 <RRSAgent>   recorded in http://www.w3.org/2002/09/30-er-irc#T11-40-44
12:39:11 <RRSAgent> ACTION: CMN make it ask the user for the URI to be tested [2]
12:39:11 <RRSAgent>   recorded in http://www.w3.org/2002/09/30-er-irc#T11-44-35
12:39:11 <RRSAgent> ACTION: CMN make it use the earl database. [3]
12:39:11 <RRSAgent>   recorded in http://www.w3.org/2002/09/30-er-irc#T11-50-15
12:39:11 <RRSAgent> ACTION: OT teach chaals PERL [4]
12:39:11 <RRSAgent>   recorded in http://www.w3.org/2002/09/30-er-irc#T12-34-38
12:39:11 <RRSAgent> ACTION: JibberJim make his bookmarklet post to http://earl.w3org/earl [5]
12:39:11 <RRSAgent>   recorded in http://www.w3.org/2002/09/30-er-irc#T12-35-04