W3C

RDF DAWG Weekly

16 Jan 2007

Agenda

See also: IRC log

Attendees

Present
LeeF, SimonR, Souri, ericP, AndyS, jeen, iv_an_ru
Regrets
PatH, EliasT
Chair
LeeF
Scribe
Souri, ericP

Contents


 

 

<LeeF> Minutes from 9 Jan: http://lists.w3.org/Archives/Public/public-rdf-dawg/2007JanMar/att-0018/09-dawg-minutes.html

<SimonR> I'll second the minutes.

<LeeF> APPROVED

<SimonR> Souri: volunteers to scribe next week.

<LeeF> Meet next 23 Jan

Lee approved the minutes (Simon seconded.).

Review ACTION Items

<LeeF> ACTION: PatH to change the entailment section around to talk about SPARQL first, then more general conditions in a normative appendix [DONE] [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action01]

<SimonR> The second paragraph of Pat's new entailment text seems (to my reading) to have two distinct topics called "instance mapping". I'd pursue it further if Pat was actually here, but it doesn't seem to make much sense to ask for an explanation until he's available. :)

<LeeF> ACTION: EricP to run the yacker tool over and annotate the existing tests [CONTINUES] [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action02]

<LeeF> ACTION: LeeF to remember that the wee, lost filter tests should be put [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action03]

<AndyS> Simon - Pat is good at replying to email.

<LeeF> CONTINUES

<SimonR> AndyS - Okay, I'll do that.

<LeeF> ACTION: LeeF to check if SteveH can eyeball Jeen's first group of tests pre-WG approval (LeeF and iv_an_ru will also try to eyeball) [CONTINUES] [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action04]

Test suite

<scribe> log: coninued action item: ACTION: EricP to run the yacker tool over and annotate the existing tests

Jeen: How do we go about approving the tests?

Lee: Approve all the syntax tests at the same time.

<ericP> are syntax-bnodes-0{3,4}.rq negative tests?

<ericP> are there any others?

<LeeF> Those are them (right now), Eric.

<AndyS> Those are the only negative ones I can find.

<LeeF> ericP: classification tool will work better if negative tests are isolated from positive tests

I think we should separate +ve and -ve syntax tests.

<AndyS> Souri - They are marked as +ve or -ve in the manifest.

<SimonR> EricP, can you just do a SPARQL query of the manifest to get yourself a list of the negative syntax tests, or would that be too circular for comfort? :)

<ericP> syntax-son-of-the-return-the-revenge-of-SPARQL2-strikes-back

<iv_an_ru> The separation is not technically needed for me but it will prevent us from discussion whether X is positive or negative. Two dirs are self-explainatory :)

<ericP> SimonR, actually, i could have algae invoke the yacker tool

<SimonR> Kurt Godel: "I am your father." DAWG: "Noooooo!"

<AndyS> If we are arranging tests by kind, (good/bad syntax) shouldn't we do that for all tests?

<LeeF> ACTION: Jeen to do further cosmetic rearranging of SyntaxDev tests and then commit them to CVS [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action05]

<ericP> scan of current /syntax/*.rq: http://lists.w3.org/Archives/Public/public-rdf-dawg/2007JanMar/att-0025/scanTests_syntax

<iv_an_ru> Naming convention instead of separate directories might be enough.

AndyS: Why do we need to separate the +ve and -ve tests?

EricP: It is easier for the tools.

LeeF: Worries whether separating the +ve and -ve tests will complicate Jeen's task a lot.

Jeen: Thinks it is not a big deal to separate them.

AndyS: Not sure about the structure (?)

<ericP> scribe: ericP

<LeeF>

LeeF: do we need to re-approve existing tests when moving them to the new testsuite?

<SimonR> We're reorganizing the tests rather than reapproving them, in my opinion.

jeen: i think they need eyeballing
... i have eyeballed triple-match/*.rq and they seem to be valid according to the current spec

<Souri> I agree with Jeen on reapproving the tests to be updated or removed.

jeen: "approval" overhead is that we need to record the meeting log in the manifests

<SimonR> The issue of approved tests that no longer reflect the current version of the language is kind of orthogonal -- even if we weren't reorganizing the tests, we'd have to do that review.

LeeF: we don't need an approval of the triple-match tests

<LeeF> ericP: thinking about process of reporting CR results

<SimonR> It's been suggested before that tests are cheap and there's no reason not to have lots of them. I'm somewhat inclined to have a "designed" collection of tests that are a roughly minimal coverage of the features. Those extra tests reduce the chance of a human ever actually reading them, which is highly desirable for correctness and understanding.

<LeeF> Simon, are you saying that it's highly desirable for humans to read them, or for humans not to read them?

<SimonR> LeeF: It's highly desirable for humans to read them.

<SimonR> LeeF: I know what I think of as a "designed" test, rather than just accumulated test data. :/

<LeeF> Simon: I understand what you're sayiing, but I worry about our ability to produce a minimal test set which is also (for some definition of the word) complete.

<AndyS> SimonR: We need to have coverage but I don't see that automatcially means remove tests we already have. More tests is better coverage , no?

<SimonR> LeeF: I think Eric's feature list gives us a notion of coverage, which will certainly help. We might look askance at any two tests with an identical set of features, for instance.

<iv_an_ru> Why not duplicate tests a bit, to have dedicated place for 5-10% of 'must read' tests and big groups of 'boring' tests. Even if readme1st tests are incomplete, running them will give an overview.

<AndyS> +1 iv_an-ru - I'm keen to help people produce better implementations. Giving them more tests is a (small) step to doing that.

<SimonR> AndyS: I'm just questioning the idea that more tests actually is helpful for implementers.

<AndyS> I'm sure the whole test suite will help improve my impls.

<AndyS> more tests = more likely to cover properly :-) I don't believe we have time to purely design a test suite.

<SimonR> Maybe I'll just stick to IRC for the nonce.

<iv_an_ru> We can't predict whether a given test is totally useless for everybody so the safe decision is to preserve tests.

<AndyS> My principle is also "least work" for Jeen

LeeF: SimonR, AndyS, iv_an_ru's discussion should go to mail or i'll bring it up in the meeting next week

rq25 status

<LeeF> current status of rq25:

<LeeF>

SimonR, re identical set of features, yeah, was gonna add that to the report

[LeeF reads 0023]

LeeF: tx for the work and update, AndyS
... Souri will mail questions about broken links

<iv_an_ru> Ups.

<AndyS> Eric: could you do the initial pub checks (is it quick and easy?) If "synatx" things are getting in the way of looking at the doc, we can remove these easily.

<LeeF> thanks, jeen

<jeen> ttyl all

<Zakim> jeen, you wanted to be excused from the meeting (i have another appointment)

<jeen> sorry to bail out early

LeeF: SimonR, do you have time and desire to review rq25 in the next 2 weeks?

<SimonR> EricP: I'll be free from about the beginning of February; you'd have to push me to get it earlier than that.

<SimonR> (My audio is cutting in and out, but I can still hear vaguely what's going on. :P)

<SimonR> O_O

<LeeF> EricP and AndyS: offering to scribe some sections rather than the whole document is also valuable

ADJOURNED

<SimonR> Adjourned at 15:26 Z.

<LeeF> ACTION: LeeF to seek early and later reviewers of rq25

Summary of Action Items

[NEW] ACTION: Jeen to do further cosmetic rearranging of SyntaxDev tests and then commit them to CVS [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action05]
[NEW] ACTION: LeeF to seek early and later reviewers of rq25 [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html]
 
[PENDING] ACTION: LeeF to remember that the wee, lost filter tests should be put [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action03]
[PENDING] ACTION: EricP to run the yacker tool over and annotate the existing tests [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action02]
[PENDING] ACTION: LeeF to check if SteveH can eyeball Jeen's first group of tests pre-WG approval (LeeF and iv_an_ru will also try to eyeball) [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action04]
 
[DONE] ACTION: PatH to change the entailment section around to talk about SPARQL first, then more general conditions in a normative appendix [recorded in http://www.w3.org/2007/01/16-dawg-minutes.html#action01]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2007/01/25 09:46:28 $