IRC log of sparql on 2012-01-24

Timestamps are in UTC.

15:00:15 [RRSAgent]
RRSAgent has joined #sparql
15:00:15 [RRSAgent]
logging to
15:00:17 [trackbot]
RRSAgent, make logs world
15:00:17 [Zakim]
Zakim has joined #sparql
15:00:19 [trackbot]
Zakim, this will be 77277
15:00:19 [Zakim]
ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start now
15:00:20 [trackbot]
Meeting: SPARQL Working Group Teleconference
15:00:21 [trackbot]
Date: 24 January 2012
15:00:27 [swh]
Zakim, who's on the call?
15:00:27 [Zakim]
SW_(SPARQL)10:00AM has not yet started, swh
15:00:28 [Zakim]
On IRC I see RRSAgent, cbuilara, swh, MacTed, AndyS, bglimm, SteveH, iv_an_ru, LeeF, Andrei, kasei, trackbot, pgearon, NickH, ya, ericP, sandro
15:00:31 [LeeF]
zakim, this will be SPARQL
15:00:31 [Zakim]
ok, LeeF, I see SW_(SPARQL)10:00AM already started
15:00:35 [Zakim]
15:00:38 [swh]
Zakim, who's on the call?
15:00:39 [Zakim]
On the phone I see ??P12, ??P13
15:00:39 [AndyS]
zakim, ??P13 is me
15:00:41 [Zakim]
+AndyS; got it
15:00:45 [swh]
Zakim, ??P12 is me
15:00:47 [Zakim]
+swh; got it
15:00:49 [Zakim]
15:00:52 [LeeF]
phone is frozen
15:00:54 [LeeF]
15:00:55 [bglimm]
Zakim, ??P19 is me
15:00:57 [Zakim]
+bglimm; got it
15:00:59 [bglimm]
Zakim, mute me
15:00:59 [Zakim]
bglimm should now be muted
15:01:02 [MattPerry]
MattPerry has joined #sparql
15:01:03 [Zakim]
15:01:21 [Zakim]
15:01:29 [LeeF]
zakkim, Lee_Feigenbaum is me
15:01:32 [LeeF]
zakim, Lee_Feigenbaum is me
15:01:35 [Zakim]
+LeeF; got it
15:01:55 [Zakim]
15:02:01 [cbuilara]
zakim, ??P29 is me
15:02:07 [Zakim]
+cbuilara; got it
15:02:09 [Zakim]
15:02:25 [LeeF]
zakim, who's on the phone?
15:02:33 [Zakim]
On the phone I see swh, AndyS, bglimm (muted), kasei, LeeF, cbuilara, MattPerry
15:02:58 [bglimm]
Zakim, unmute me
15:03:09 [AndyS]
scribenick: AndyS
15:03:13 [Zakim]
bglimm should no longer be muted
15:03:17 [AndyS]
scribe: Andy Seaborne
15:03:21 [chimezie]
chimezie has joined #sparql
15:03:41 [LeeF]
15:03:45 [Zakim]
+ +1.540.841.aaaa
15:03:47 [LeeF]
topic: Admin
15:04:10 [LeeF]
PROPOSED: Approve last week's minutes from
15:04:17 [pgearon]
Zakim, aaaa is me
15:04:17 [Zakim]
+pgearon; got it
15:04:27 [LeeF]
Regrets: Axel, Olivier
15:04:39 [chimezie]
chimezie has joined #sparql
15:04:40 [LeeF]
RESOLVED: Approve last week's minutes from
15:04:49 [chimezie]
Zakim, what is the passcode?
15:04:52 [Zakim]
the conference code is 77277 (tel:+1.617.761.6200, chimezie
15:04:53 [LeeF]
Next meeting is on 1/31
15:05:28 [Zakim]
+ +1.216.368.aabb
15:05:38 [chimezie]
Zakim +1.216.368.aabb is me
15:05:51 [chimezie]
Zakim, mute me
15:05:51 [Zakim]
sorry, chimezie, I do not know which phone connection belongs to you
15:05:58 [LeeF]
Lee: RDF WG?
15:06:05 [LeeF]
AndyS: pretty quiet, discussing graph use cases
15:06:11 [LeeF]
...small point around turtle and hex binary
15:06:19 [chimezie]
Zakim, 1.216.368.aabb is me
15:06:19 [Zakim]
sorry, chimezie, I do not recognize a party named '1.216.368.aabb'
15:06:22 [LeeF]
...about whitespace, but I don't think there will be progress on that
15:06:50 [AndyS]
topic: tests
15:07:10 [LeeF]
implementation report is here:
15:08:07 [chimezie]
Zakim, +1.216.368.aabb is me
15:08:07 [Zakim]
+chimezie; got it
15:08:12 [chimezie]
Zakim, mute me
15:08:12 [Zakim]
chimezie should now be muted
15:08:40 [AndyS]
kasei:chnages Update syntax shortcuts, qname escapes, a few other things
15:08:50 [LeeF]
zakim, who's on the phone?
15:08:50 [Zakim]
On the phone I see swh, AndyS, bglimm, kasei, LeeF, cbuilara, MattPerry, pgearon, chimezie (muted)
15:08:53 [AndyS]
... last run Oct 2011.
15:09:03 [AndyS]
... despite CVS timestamp.
15:09:20 [chimezie]
I can do all of the above but just for some entailment tests
15:09:20 [AndyS]
Leef: anyone else in a position to run the test suite?
15:09:24 [pgearon]
not me at all
15:09:35 [MattPerry]
We use ARQ, so our results will basically be the same
15:09:43 [bglimm]
Hopefully I can run the tests in a few weeks (entailment tests mainly though)
15:10:30 [AndyS]
LeeF:want to batch approve all tests in two impls.
15:11:19 [bglimm]
Michael Schneider wanted to implement ent. regimes. I haven't talked to him recently to see where he stands
15:11:24 [AndyS]
AndyS: Reach out to other implementations.
15:11:45 [AndyS]
LeeF: Will go outbound soon so we get feedback on mechanisms ASAP.
15:12:05 [LeeF]
ACTION: Lee to post about test suite on public-sparql-dev and -comments list and reach out to implementors
15:12:05 [trackbot]
Created ACTION-579 - Post about test suite on public-sparql-dev and -comments list and reach out to implementors [on Lee Feigenbaum - due 2012-01-31].
15:12:29 [AndyS]
LeeF: Query first
15:13:33 [AndyS]
Mechanism -- approve all but specific tests based on last report.
15:14:08 [LeeF]
15:16:29 [AndyS]
ARQ runs the agg tests and functions all OK (from CVS)
15:19:16 [AndyS]
ARQ runs the update silent test all successesfully
15:19:35 [LeeF]
close ACTION-575
15:19:35 [trackbot]
ACTION-575 Propose tests for ADD, MOVE, and COPY closed
15:19:50 [AndyS]
LeeF: newer tests from Matt. ARQ reports running them OK.
15:19:56 [LeeF]
close ACTION-574
15:19:56 [trackbot]
ACTION-574 Propose tests for positive use of EXISTS closed
15:20:15 [AndyS]
.. EXISTS tests look OK (2 impls pass the tests)
15:21:16 [LeeF]
ACTION: Greg to re-run the implementation report and check in an updated one
15:21:17 [trackbot]
Created ACTION-580 - Re-run the implementation report and check in an updated one [on Gregory Williams - due 2012-01-31].
15:22:59 [kasei]
15:23:02 [LeeF]
ack kasei
15:23:02 [AndyS]
LeeF: For CR - snapshot place and fixed report place
15:24:55 [AndyS]
kasei: SHA384 still there
15:25:00 [AndyS]
AndyS: I'll remove them.
15:25:05 [AndyS]
topic: protocol
15:26:23 [AndyS]
LeeF: Easier to start again - need an online service that will poke a server with the protocol.
15:26:33 [AndyS]
... service description testing ...
15:27:05 [AndyS]
kasei: Online form, for a URL, get back a report of SD conformance 9(3 tests currently)
15:27:11 [kasei]
15:27:30 [AndyS]
... conneg -> EARl if RDF.
15:28:38 [AndyS]
LeeF: coverage page says we will have 2+ impls. We may need to encourage people to put up SD's
15:29:50 [AndyS]
... mildly related ... anyone at SemTech presenting SPARQL 1.1?
15:30:18 [AndyS]
LeeF: want similar set up for protocol - something that poke a service.
15:30:38 [AndyS]
.. drive by the examples in the protocol doc.
15:30:53 [AndyS]
... one new thing - query as application/sparql-query
15:31:50 [LeeF]
ACTION: Lee to work changes to protocol TR publication back into editor's draft
15:31:50 [trackbot]
Created ACTION-581 - Work changes to protocol TR publication back into editor's draft [on Lee Feigenbaum - due 2012-01-31].
15:32:35 [AndyS]
LeeF: Overall next week, approve all that is green x2 impls
15:33:04 [LeeF]
15:33:04 [trackbot]
ACTION-580 -- Gregory Williams to re-run the implementation report and check in an updated one -- due 2012-01-31 -- OPEN
15:33:04 [trackbot]
15:33:11 [AndyS]
Topic: TSV/CSV formats
15:35:16 [LeeF]
15:36:40 [AndyS]
LeeF: Concern is that it is two different formats.
15:37:07 [AndyS]
... CSV is string forms, TSV is full Turtle terms.
15:37:10 [swh]
15:37:32 [AndyS]
... found this strange - not the way I'd do it.
15:37:57 [AndyS]
... but not an issue to hold up over. Still want some discussion on it though.
15:38:05 [LeeF]
ack swh
15:39:13 [AndyS]
swh: Understand the concern; CSV in excel often does not want Turtle details (experience); TSV is for programs. No user feedback againt it.
15:39:17 [AndyS]
15:39:22 [LeeF]
ack Andrei
15:39:25 [LeeF]
ack AndyS
15:39:37 [swh]
AndyS: experience is that TSV is quite popular
15:39:46 [swh]
… people use CSV but a little bit
15:39:53 [swh]
… and it's very fast to produce
15:40:10 [swh]
… get the why does ntriples parse faster than turtle effect
15:40:18 [swh]
… don't see much use of CSV
15:40:31 [swh]
… people dont see it in the wild as being different formats, lots of confusion
15:40:40 [swh]
… there's a lot of confusion in the wild
15:41:09 [AndyS]
LeeF: can be explained
15:41:35 [swh]
AndyS: which ones does Anzo use
15:41:36 [AndyS]
... Anzo does the string form (CSV)
15:42:39 [kasei]
I'm in agreement with Lee. Seems weird, but think it can be explained.
15:43:23 [AndyS]
LeeF: Consensus on two formats - issue of confusion recognized as possible.
15:44:03 [AndyS]
AndyS: Needs more examples
15:44:37 [AndyS]
LeeF: Seems better to make it REC track.
15:44:58 [AndyS]
... next step - complete doc and then move forward through process.
15:45:36 [AndyS]
Topic: AOB
15:45:44 [chimezie]
Zakim, unmute me
15:45:44 [Zakim]
chimezie should no longer be muted
15:46:00 [AndyS]
LeeF: Chime - anything about graph store?
15:46:00 [kasei]
q+ to ask about ACTION-573 (aggregate errors)
15:46:13 [AndyS]
chime: next time please
15:46:23 [LeeF]
ack kasei
15:46:23 [Zakim]
kasei, you wanted to ask about ACTION-573 (aggregate errors)
15:47:02 [AndyS]
kasei: need more agg-error tests?
15:47:06 [LeeF]
close ACTION-573
15:47:06 [trackbot]
ACTION-573 Propose tests for aggregate evaluation errors closed
15:47:07 [AndyS]
LeeF: No - it's done.
15:47:48 [AndyS]
15:47:50 [Zakim]
15:47:52 [Zakim]
15:47:53 [Zakim]
15:47:54 [AndyS]
tarck, make logs public
15:47:55 [MattPerry]
15:47:57 [bglimm]
15:48:02 [AndyS]
trackbot, make logs public
15:48:02 [trackbot]
Sorry, AndyS, I don't understand 'trackbot, make logs public'. Please refer to for help
15:48:04 [Zakim]
15:48:06 [Zakim]
15:48:06 [Zakim]
15:48:11 [AndyS]
rrsagent, make logs public
15:48:12 [Zakim]
15:48:13 [LeeF]
15:48:13 [trackbot]
ACTION-574 -- Gregory Williams to propose tests for positive use of EXISTS -- due 2012-01-24 -- CLOSED
15:48:13 [trackbot]
15:48:18 [LeeF]
15:48:18 [trackbot]
ACTION-546 -- Gregory Williams to review query 2nd LC as soon as it's ready -- due 2011-11-15 -- OPEN
15:48:18 [trackbot]
15:48:23 [LeeF]
close ACTION-546
15:48:23 [trackbot]
ACTION-546 Review query 2nd LC as soon as it's ready closed
15:48:32 [Zakim]
15:48:37 [kasei]
LeeF, I thought you had something for me at the end of the call?
15:48:44 [Zakim]
15:48:44 [kasei]
protocol or SD related?
15:48:45 [Zakim]
SW_(SPARQL)10:00AM has ended
15:48:46 [Zakim]
Attendees were AndyS, swh, bglimm, kasei, LeeF, cbuilara, MattPerry, +1.540.841.aaaa, pgearon, chimezie
15:57:22 [kasei]
AndyS, you pass the new COPY tests?
15:58:49 [LeeF]
what did I have for you?
15:59:06 [kasei]
I don't think the error condition text in the Update spec say enough to imply the results for copy05.
15:59:19 [kasei]
LeeF, I'm not sure. I just thought you said you were going to stick around to discuss something.
16:00:51 [kasei]
16:01:33 [kasei]
Update says "Each request SHOULD be treated atomically by a SPARQL 1.1 Update service" but then the definition says "An Update Operation Op is an atomic operation..."
16:01:52 [kasei]
e.g. the SHOULD-hedge isn't present in the actual definition.
16:02:34 [kasei]
which is a big problem for systems that don't have support for atomic transactions in the underlying graphstore.
16:07:40 [LeeF]
hang on, kasei
16:07:46 [LeeF]
the first thing you quoted is talking about REQUESTS
16:07:51 [LeeF]
the second thing is talking about OPERATIONS
16:07:55 [LeeF]
big difference, right?
16:08:15 [kasei]
ah, maybe
16:08:29 [Andrei]
Andrei has joined #sparql
16:08:30 [kasei]
in that case, the problem is just with the new test cases
16:08:36 [LeeF]
don't know what the impact is for copy05 :)
16:09:00 [kasei]
16:09:31 [kasei]
how does the request/operation distinction affect the shortcut syntaxes for things like COPY? Is COPY one operation, or two?
16:10:03 [kasei]
it says it's "equivalent to" a syntax form with two operations.
16:10:58 [LeeF]
very interesting question
16:11:06 [kasei]
and since the first operation is a DROP, whether it's atomic or not (when a problem is encountered during the INSERT) is a big deal.
16:11:16 [LeeF]
I'm going to hesitate to say "i told you so" about prematuring adding shortcuts, but........ ;-)
16:11:27 [LeeF]
definitely warrants an email to the group
16:11:46 [kasei]
don't point that at me! I don't think I was promoting these shortcuts :\
16:11:59 [kasei]
yeah, i'll write it up and send to the group.
16:13:05 [kasei]
but I may be the most affected by this issue. Not sure how many systems are non-transactional under the hood.
16:13:14 [LeeF]
no, you weren't, it was a generic "you" :-)
16:13:23 [kasei]
16:14:36 [Andrei]
Andrei has joined #sparql
16:24:55 [AndyS]
Yes - ARQ passes the COPY tests.
16:27:18 [AndyS]
Not clear why that text is relevant - test is sequential. requests are atomic .. not operations ... otherwise madness occurs.
16:29:46 [kasei]
AndyS, so you're saying COPY is two operations, not one?
16:30:03 [kasei]
because it's referred to as "the COPY operation".
16:30:21 [kasei]
a bigger problem is that it isn't treated the same as its "equivalent" expansion.
16:30:29 [AndyS]
I'm saying it makes no difference although for completely different reasons it had better be one.
16:30:46 [AndyS]
request are atomic, not operations.
16:31:08 [kasei]
no, the doc explicitly says the reverse. operations are atomic. requests are RECOMMENDED to be atomic.
16:31:26 [Andrei]
Andrei has joined #sparql
16:32:46 [kasei]
given the current wording, I think whether the shortcuts are treated as one or multiple operations makes a very big difference.
16:32:54 [AndyS]
Needs fixing. But I think the intention is that defn atmic request is saying these are the building blocks. Different sense of "atomic"
16:33:36 [kasei]
ok, but the text about atomic requests is only a SHOULD.
16:33:46 [kasei]
i'm going to have big problems if it's a MUST.
16:40:19 [AndyS]
ARQ can't do MUST for all storages. TDB is now transaction safe; memory storage does pseudo-transactions by locking which is indistinguishable. Failure recovery for in-memory is easy to implement :-)
16:42:33 [kasei]
ok. for the stores that aren't transaction safe, though, isn't this issue relevant?
18:04:27 [Zakim]
Zakim has left #sparql
19:03:43 [swh]
swh has joined #sparql
19:25:37 [AndyS]
kasei, EARL results for ARQ: CSV checking manual, doign bNode isomorphism checking just for CSV is a not a reasonable use of time.
19:26:41 [AndyS]
kasei, re transactions, if not transaction safe, and if atomic, no issue - it's the request that needs to be made exclusive. i.e. the HTTP request.
19:50:27 [Andrei]
Andrei has joined #sparql
19:55:36 [kasei]
AndyS: what I'm suggesting is that a multi-op request could run some of the ops, and fail in the middle of the reqest, and leave the graphstore in a state that isn't the requests expected pre- or post- state.
19:56:37 [AndyS]
I see that ... but you said it wasn't transactional ... so it may happen even in a single operation not being truely atomic.
20:00:50 [kasei]
ah, sure. I was considering that a separate issue.
20:01:14 [kasei]
since you could have a system where ops actually are atomic, but because of these multi-op requests, you still get the inconsistent state.
20:01:23 [kasei]
also, that new EARL looks good, thanks.
20:05:06 [kasei]
generating a new implementations report now...
20:47:21 [kasei]
new implementations report checked in:
21:15:04 [AndyS]
kasei - cool ... and good night.
21:35:00 [Andrei]
Andrei has joined #sparql
21:38:03 [kasei]
is there an appropriate property to relate an endpoint URI with the software that sits behind it?
21:38:07 [kasei]
<> ex:??? <url-for-virtuoso>
22:00:31 [Andrei]
Andrei has joined #sparql
22:37:44 [pgearon]
pgearon has joined #sparql
22:44:54 [Andrei]
Andrei has joined #sparql
23:08:34 [Andrei]
Andrei has joined #sparql