Log of the #fcrepo channel on chat.freenode.net

Using timezone: Eastern Standard Time
* jmignault leaves02:53
* dwilcox joins08:12
* jonroby joins08:45
* whikloj joins08:57
* acoburn joins09:07
* awead joins09:09
* dhlamb joins09:23
<jonroby>anyone know how to make mockito work nicely when you generate javadocs?09:34
<awoods>jonroby: An example of the issue you are hitting would be helpful.09:35
<jonroby>awoods: https://gist.github.com/robyj/56aaa2a19edc0267885f09:37
<awoods>jonroby: Do you think those errors are related to Mockito?09:38
<jonroby>awoods: those line numbers come after mockito tags09:39
<awoods>jonroby: which command are you running to get those errors?09:40
<jonroby>awoods: mvn javadoc:test-aggregate
* osmandin joins09:43
* ajs6f joins09:45
* ksclarke joins09:48
<awoods>jonroby: It looks like a bug in the version of the javadoc.plugin. Bump the version in the following line to 2.10.309:55
jonroby: If that works, could you please make sure we are consistently on the same version of the javadoc plugin in the other modules you are fixing?09:56
<jonroby>awoods: will do
awoods: thanks, checking it works now
awoods: it works, just some @return warnings to fix. cheers!09:57
<awoods>jonroby: great
* github-ff joins10:00
[fcrepo4] yinlinchen closed pull request #799: Remove UUID from default transform (master...fcrepo-1552) http://git.io/vTozW
* github-ff leaves
* cherveny leaves10:03
* github-ff joins10:05
[fcrepo4] yinlinchen opened pull request #804: remove fedora:uuid (master...fcrepo-1552) http://git.io/vkn2A
* github-ff leaves
<dhlamb>does fcrepo accept sparql in the DELETE WHERE {} INSERT DATA {} format? or is it only DELETE {} INSERT {} WHERE {} ?10:16
<ajs6f>dhlamb: It should do whatever Jena can do.
<dhlamb>ajs6f: ok, thx. i'm running into an issue with it not liking what Jena spits out, and was wondering if i hit a bug or am just doing something wrong.10:17
ajs6f: sounds like i'm just doing something wrong ><10:18
<ajs6f>dhlamb: Not liking what Jena spits out? Can you unpack that a bit?
<acoburn>dhlamb: did you include a semicolon after the DELETE WHERE {} clause?
<dhlamb>ajs6f: acoburn: relevant gist: https://gist.github.com/daniel-dgi/ddc3e32b715bb25b74b910:20
i'm trying to update values, and instead of getting replaced, they're simply getting appended10:21
<ajs6f>acoburn: Is it a Java system property for the location of the Fuskei config file, or something else?
dhlamb: This may have nothing to do with it, but you really can't delete or add things like "rdf:type ldp:RDFSource".10:22
<acoburn>ajs6f: fuseki2?
<dhlamb>ajs6f: acoburn: sry for all the clutter on that query, but it's dc:title and a few other properties that don't seem to be replacing. i mangled it into DELETE {} INSERT {} WHERE {} format and it seemed to work
<ajs6f>acoburn: Yep.
<acoburn>ajs6f: I assume you're using the war deployment (not standalone)10:23
<ajs6f>acoburn: It's not me, actually, but yes, that's what's happening.
<acoburn>ajs6f: there was an issue with this that was recently fixed in trunk
<ajs6f>acoburn: Finding a config file?10:24
<acoburn>ajs6f: but I don't think it's made it into the current release
ajs6f: yes
<ajs6f>acoburn: Crum. Is it possible at all?
<whikloj>awoods: so those LDP tests are failing because the test-suite trys to GET the uri.
* github-ff joins
[fcrepo4] yinlinchen opened pull request #805: Replace fedora:Blanknode with fedora:Skolem in the code and in the on… (master...fcrepo-1394) http://git.io/vknPg
* github-ff leaves
<whikloj>awoods: at least 2 of the 3 are, the other is an actual failure10:25
<acoburn>ajs6f: yes, it's possible, but a bit "hackish"
ajs6f: I have to look back through my notes
<awoods>whikloj: are you saying the LDP tests are flawed?10:26
<whikloj>awoods: either that or a Data URI is not valid for a constrainedBy Link header10:27
* travis-ci joins
robyj/fcrepo4#2 (fcrepo-1556 - a0854c7 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo4/compare/e60a1c2f717c...a0854c75d00f
Build details : http://travis-ci.org/robyj/fcrepo4/builds/64425211
* travis-ci leaves
<ajs6f>acoburn: Thanks for any advice.10:28
awoods:whikloj: Or there's an assumption in LDP that hasn't been properly expressed in the spec.
<whikloj>ajs6f/awoods: that is possible, maybe cbeer was the first to think of using a data URI?10:29
<awoods>whikloj: it would be interesting to see if you hard-coded a URL into the MalformedRdfExceptionMapper it the test would be happier.10:30
<whikloj>awoods: I was just going to try that, I'll let you know.
awoods/ajs6f: This is the function that kills us. https://github.com/w3c/ldp-testsuite/blob/6831c3dbd30461f28f94f26a4a0c89ddcb80c3b8/src/main/java/org/w3/ldp/testsuite/test/RdfSourceTest.java#L493-L50510:31
* travis-ci joins
robyj/fcrepo-build-tools#3 (fcrepo-1556 - 6a562cb : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-build-tools/compare/0b2d090b990e...6a562cbcb36c
Build details : http://travis-ci.org/robyj/fcrepo-build-tools/builds/64427893
* travis-ci leaves
<ajs6f>acoburn: Was it this stuff:10:39
<acoburn>ajs6f: that might be part of it10:40
ajs6f: the issue was that you can drop a configuration file in the "config" directory (or some such location)
ajs6f: but then, when fuseki tries to read the file, it would have the wrong cwd and so throw an exception10:41
<jonroby>awoods: committer call today?10:48
<awoods>jonroby: every Thursday... like a ray of sunshine. https://wiki.duraspace.org/display/FF/2015-05-28+-+Fedora+Tech+Meeting10:49
<acoburn>ajs6f: I'm not getting fuseki to load custom configuration files when deployed as a war
ajs6f: I thought that, by creating a symlink in a particular directory would get around the bug, but that doesn't seem to work10:50
<ajs6f>acoburn: Just so you know, I'm asking because I'm trying to help someone in #jena.
<acoburn>ajs6f: the easy answer is to just modify the built-in config.ttl10:51
ajs6f: it should be in $FUSEKI_BASE/config.ttl
<ajs6f>acoburn: Okay, cool. I'll so advise.10:52
<whikloj>acoburn: is that for fuseki 2.0?
<f4jenkins>Yippee, build fixed!
Project fcrepo4-T2 build #249: FIXED in 4 min 30 sec: http://jenkins.fcrepo.org/job/fcrepo4-T2/249/
* awoods: Update maven site related versions in response to:
* awoods: Change MalformedRdfException to a RuntimeException
<acoburn>whikloj: yes, when deployed as a war
<whikloj>acoburn: I tried editing the config.ttl to add a default dataset, but it still didn't create it. I thought it was a bug, I'll have to try again.10:53
* travis-ci joins10:54
robyj/fcrepo-message-consumer#6 (fcrepo-1556 - 4299873 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-message-consumer/compare/d412e3bfdc34...429987362d9c
Build details : http://travis-ci.org/robyj/fcrepo-message-consumer/builds/64429754
* travis-ci leaves
<whikloj>awoods: replacing the data uri with http://www.google.ca cleared those two errors, now just one error10:59
<acoburn>whikloj: I just edited the config.ttl file and it worked fine with fuseki2
* travis-ci joins11:00
robyj/fcrepo-module-auth-rbacl#2 (fcrepo-1556 - 7e20b86 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-module-auth-rbacl/compare/7d43add359d4...7e20b86c4a32
Build details : http://travis-ci.org/robyj/fcrepo-module-auth-rbacl/builds/64430837
* travis-ci leaves
<awoods>whikloj: that is interesting... I wonder if the test should allow data-URIs?
<whikloj>acoburn: my config must be wrong
<awoods>whikloj: maybe a question for the LDP Working Group.
<whikloj>awoods: if we knew someone on the LDP WG
<awoods>whikloj: we are in luck11:01
<acoburn>whikloj: nothing fancy, but this works: https://gist.github.com/acoburn/9c64c505e90c5c5d5776
<whikloj>acoburn: thanks11:02
* dshalvi joins11:04
* travis-ci joins11:08
robyj/fcrepo-module-auth-xacml#2 (fcrepo-1556 - 9a22afb : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-module-auth-xacml/compare/c69b94a8afcd...9a22afbdd1fb
Build details : http://travis-ci.org/robyj/fcrepo-module-auth-xacml/builds/64432456
* travis-ci leaves
<dhlamb>and we just lost awoods11:24
<jonroby>aaaaand we're done...good meeting :)
<escowles>anyway, +1 to setting up a central RDF file that describes our constraints
<ajs6f>We just lost the call, because THE CHAIRPERSON HAS DISCONNECTED.
<osmandin>no more note taking for me today
<awoods>all... come on back11:25
<ajs6f>escowles: So, you're cool to require network to run tests?
* travis-ci joins
robyj/fcrepo-camel#1 (fcrepo-1556 - caa6766 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-camel/commit/caa6766ff8fe
Build details : http://travis-ci.org/robyj/fcrepo-camel/builds/64436061
* travis-ci leaves
<whikloj>escowles: Could we link to a specific part of the RDF file of constraints?
<escowles>ajs6f: yes -- it doesn't seem like unreasonable requirement to me
<ajs6f>whikloj: We might could, or we could use multiple files.
escowles: I'm thinking about working while traveling, but that's really my problem.11:26
<whikloj>awoods has the call back up
<awoods>whikloj: yes
<escowles>whikloj: ajs6f: yes, either multiple files or hash anchors: fedora.info/constraints#invalidSubjectDomain
<whikloj>ajs6f: ok, cause I think we want to be specific with the constraints for that header11:27
<escowles>ajs6f: yes, having it be served by the repo could also work
<ajs6f>escowles: I'm a little cautious about hash anchors because I don't understand what they do when you ask for Ntriples or turtle.
<terrellt>Typically you'd get all of fedora.info/constraints and it'd be up to you to filter by subject.11:28
<ajs6f>terrelt: Yes, but we want to also provide human-readable output.
<whikloj>We might need to provide some code so they know (or the machine knows) what constraint to look at.
<terrellt>HTML serializations usually jump to the appropriate anchor, are NT and TTL human-readable?
I mean, -technically-, but I've never seen an NT serialization with anchors do anything but return the whole graph when you request it and expect you to jump appropriately.11:30
<escowles>terrellt: dunno -- all of the RDF serializations have issues -- but we could easily transform it to a human-readable HTML page
<ajs6f>terrellt: A) Yes, they are. B) It's okay to throw everything back now and have people filter, but I don't like the pattern that sets. With more possible faults, that would become burdensome.
* github-ff joins11:31
[fcrepo-camel] robyj opened pull request #71: - bumped javadoc plugin version in pom.xml to 2.10.3 (master...fcrepo-1556) http://git.io/vkc3T
* github-ff leaves
<terrellt>That being said, I've always liked explicit URIs - fedora.info/constraints/invalidSubjectDomain
<ajs6f>terrellt: +1 to that.
<terrellt>So I want to make a ticket for this, but I'm not totally sure how to explain it, so I'm going to try here to boil it down.11:32
And if I say anything that's blatantly false, please correct me
When I POST a resource and some triples to Fedora it creates a resource and adds the appropriate server-managed triples. Fedora knows how to separate these, and you can filter them out with a header. Which is great. When I GET a resource without the header, I get the server managed triples alongside my triples. In ActiveFedora I WANT the server managed triples for system reasons - solrizing when the resource was created so11:36
I can sort on it, for example. However, I DON'T want to make those triples editable - when you persist a change to those triples Fedora gets upset, and there's no reason I can think of for the user to mess with them
But I have no good way to filter between the server managed triples and the ones I persisted (for example, we persist RDF.type statements, and separating ours from Fedora's is inefficient) WITHOUT sending two GETs - one asking for server managed triples and one without.11:37
<terrellt>Related, if I PUT a resource with no RDF.type statements, Fedora will get pissed because I've removed Server Managed Triples. That's not the end of the world - I can handle that locally if I can separate SMT from my triples, or I can just always use SPARQL update.11:39
Assuming what I've said is true, and I haven't missed out something huge, I propose two things: A way to request a serialization that supports named graphs and puts server managed triples in one (DO NOT have to support user-created named graphs), and make it so when I PUT it adds those triples to the default graph, and doesn't mess with SMT.11:40
The second's not necessary, but the first would be lovely
<ajs6f>terrrellt: Would you be happy if you just didn't have to fool with the type triples? If you could just leave them out?11:41
<terrellt>What do you mean?
<ajs6f>terrellt: If you could leave them off, PUT your PUT, and it would be accepted?
* github-ff joins
[ontology] robyj opened pull request #27: - added javadoc plugin version 2.10.3 to pom.xml (master...fcrepo-1556) http://git.io/vkcn7
* github-ff leaves
<terrellt>Yes, that's the second part.
<acoburn>terrellt: could you use an external 3store for this?11:42
<dhlamb>i have excised myself from the list :(
<terrellt>acoburn: I'm not sure why I would? Unless I misunderstand
* travis-ci joins
robyj/fcrepo-audit#5 (fcrepo-1556 - 6f8bee4 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-audit/compare/b394427c0c35...6f8bee439d3a
Build details : http://travis-ci.org/robyj/fcrepo-audit/builds/64438256
* travis-ci leaves
<ajs6f>awoods: Is AB coming?
<acoburn>terrellt: in order to separate triples into particular named graphs, then you could make a single request11:44
<whikloj>ajs6f: you need the agenda? https://wiki.duraspace.org/display/FF/2015-05-28+-+Fedora+Tech+Meeting
<ajs6f>whikloj: No, thanks— I just really like it. AGENDA++11:45
<whikloj>ajs6f: haha
* scossu joins
<terrellt>acoburn: I'd still have to do two requests to get the server triples and my triples, right? At which point I don't know what another 3store would do for me11:46
<ajs6f>awoods: Where is that strawman?
terrelt: Is this about doing something you can't, or about doing something you can do more efficiently?11:47
<terrellt>Something I can do more efficiently. We already have a request-count problem, another GET for every resource is painful (and requires a pretty large amount of refactoring in ActiveFedora, but that's our problem)11:49
<acoburn>terrellt: why not cache the requests? or have an intermediate ETL system to reduce the request count?11:50
<ajs6f>terrellt: What is "a request-count problem"? Do you have actual numbers to reach, or a vague and unpleasant sense of unease or nausea?11:51
<acoburn>terrellt: just embrace it and call it a "gossip protocol"11:52
<terrellt>I can make up arbitrary numbers for situations we have, but that's not useful. The crux of this is I can't, in one request, differentiate between what the server added and what I did. (By the way, can I request JUST the server triples?)11:53
And I think it would be useful if I could.11:54
<ajs6f>terrellt: I don't think you can get just the server triples, but you can file a ticket for that. I don't know easy that would be impl-wise, but you might as well ask.
<whikloj>ajs6f: I think barminator did not have his strawman ready to publish, discussed it https://wiki.duraspace.org/display/FF/2015-04-23+-+Fedora+Tech+Meeting
<ajs6f>whikloj: Thanks!
<dhlamb>sry gtg12:01
<awoods>osmandin: please let me know when you are done with the meeting notes. Thanks!12:04
<osmandin>awoods: sure
<awoods>terrellt: it seems like you should be able to only get ServerManaged: https://wiki.duraspace.org/display/FEDORA4x/RESTful+HTTP+API+-+Containers#RESTfulHTTPAPI-Containers-GETRetrievethecontentoftheresource12:18
terrellt: "include=ServerManaged; omit=others"12:19
* osmandin leaves12:21
Using the documentation. Clever.12:22
<awoods>ajs6f: it is a trick I use occasionally.
<ajs6f>One clever trick to avoid confusion with your API. Doctors hate it!12:24
* osmandin joins12:32
<whikloj>awoods/ajs6f: re:FCREPO-1411 tests still fail, seems that the system doesn't care that the object doesn't exist as long as the id convertor can transform it.12:34
* github-ff joins12:35
[fcrepo4-client] robyj opened pull request #21: - added javadoc plugin to pom.xml, set to version 2.10.3 (master...fcrepo-1556) http://git.io/vkc16
* github-ff leaves
<ajs6f>whikloj: "the object"?
<whikloj>ajs6f: sorry object == subject
<ajs6f>whikloj: So you're saying that RDF that has an in-domain subject that doesn't actually exist passes when it shouldn't?12:36
<whikloj>ajs6f: I'm not saying it shouldn't ('cause I don't know the spec very well). But yes, using your IT test it allows it to apply RDF to a subject that doesn't exist12:37
<ajs6f>whikloj: Does that code get used to persist RDF about _new_ resources? (Which genuinely don't exist, _yet_.)
<whikloj>ajs6f: It appeared to, but I'll confirm.12:38
<ajs6f>whikloj: If it be so, we may have to be more subtle about the test. We may have to apply different tests for new vs. extant resources.12:39
whikloj: Or we may have to rearrange the worlkflow so that mutations are only applied to extant resources?12:40
<osmandin>awoods: Wiki updated with minutes.12:42
* osmandin leaves12:43
<whikloj>ajs6f: actually I think the test might be wrong12:44
<whikloj>both URIs are built using the server address and a getRandomUniquePid, so when the PUT executes it just creates the second one. Is the property supposed to be read only?12:46
<ajs6f>whikloj: What property?
<whikloj>ajs6f: <info:rubydora#label>12:47
<ajs6f>whikloj: I'm totally lost. Can you give me a pointer into the code, so we can look at it together?
<whikloj>ajs6f: forget it I think I misunderstood the test
<ajs6f>whikloj: Okay, that works for me.
<whikloj>ajs6f: point is I wanted to see if FCREPO-1411 was changed by completing FCREPO-1417. It is not.12:48
<ajs6f>whikloj: Okay, as long as you're cool. I feel like 1411 is going to resolve almost immedaitely after 1417, no?12:50
<whikloj>ajs6f: not if the point of 1411 was that you can't apply RDF for an object that is not the endpoint of your request.12:51
ajs6f: your test creates an object and creates the path to a second object, then applies RDF for both. You were asserting this should fail. But it just creates the missing object because it is "inDomain"12:52
<ajs6f>whikloj: It _creates_ the missing resource? (P.S. let's not use the term "object" except for RDF-objects, okay?)12:53
<whikloj>ajs6f: sorry, yes it creates the missing resource
<ajs6f>whikloj: That is definitely a flat-out bug. It should never be creating resources unless A) they are hash-URI things, or B) they are blank nodes. And my test _did_ fail when I gave it osmandin, so something has changed since then.12:54
<whikloj>ajs6f: you misunderstand me, it still fails. So I don't think 1417 will affect it at all.12:55
<ajs6f>whikloj: OOOOOH. Okay, but _how_ does it fail? IOW what kind of response does the user get? It's the kind of response that I suspected might be affected by 1417.12:56
<whikloj>ajs6f: See the action does not fail, so the client gets 201. Expected 403
<ajs6f>whikloj: Sure, but does that 403 have an appropriate Constraint: header? Does that header have a good URI as value? Is the message as we would expect?12:58
<whikloj>ajs6f: they get a 201
<ajs6f>whikloj: Sorry, yes, but after 1411 is done, _will_ they get the right thing? Or does 1411 need to be rewritten to explain what the right thing is and how to use the machinery out of 1417 to do it?12:59
<whikloj>ajs6f: ah gotcha. we will have to see.13:00
<ajs6f>whikloj: right, it's not a question we can answer until the end of 1417.
whikloj: I wouldn't even worry any more about 1411 until you feel like you're really happy with 1417.
<whikloj>ajs6f: I guess my question is how do we define a resource as being "out-of-domain"?13:01
ajs6f: I assumed it would have a different server address, is that correct?
<ajs6f>whikloj: Let me back you up there a sec: what is the purpose of the test of in-domainness? IOW when are you using that test?13:02
<whikloj>ajs6f: I was running this test. https://github.com/osmandin/fcrepo4/blob/1411/fcrepo-http-api/src/test/java/org/fcrepo/integration/http/api/FedoraLdpIT.java#L766-L78213:03
<awoods>jonroby: do you have a PR for fcrepo-audit?13:06
<ajs6f>whikloj: Sorry, I mean that we're using the quality of in-domainness as a test, a unary predicate, to accept certain URIs and not others. That's the sense of "test" I meant. With that in mind, when (in workflow) / where (in code) are we using it, is the thing I'm trying to make sure we've thought about.
In other news, the simple answer to your question is "We don't, in the code on which you are working. The implementors of IdentifierConverter are responsible for that. That's why the test/predicate is a method on IdentifierConverter.
<awoods>jonroby: the commit you list in the ticket is not a PR: https://jira.duraspace.org/browse/FCREPO-155613:07
<whikloj>ajs6f: right so I guess my general question is (as you said before) should we have a more comprehensive test?13:08
<ajs6f>whikloj: I'm just not sure how this is arising for your issue. You should be able to just accept an IdentifierConverter and rely on whatever test it supplies. If you can't, it implies that there exists a problem with our contracts, right?13:09
* github-ff joins13:10
[fcrepo-audit] awoods created fcrepo-1556.test (+1 new commit): http://git.io/vkcpg
fcrepo-audit/fcrepo-1556.test 34ee563 Andrew Woods: test, then remove this branch
* github-ff leaves
<whikloj>ajs6f: because I wanted to see if my code for 1417 would affect the results of your tests for 1411. This is not affecting my work. We can let it die here. :)
<ajs6f>whikloj: Okay, sure. I feel like this is such a subtle thing that we're not going to be able to figure it out via IRC. Definitely let's let it relax and you ping me if you want.13:11
* osmandin joins13:18
<dhlamb>w/r/t what i brought up earlier. now that it's not so mired in context, it does look like i've stumbled upon a bug. can someone else try this out?13:39
looks like when you add a new term and update an existing one, the update becomes an addition and you get two terms
<ajs6f>dhlamb; What are you runing these against? How did you set p the original resource?
<dhlamb>ajs6f: through the UI13:40
ajs6f: and also through fcrepo-camel, but confirmed through UI
<ajs6f>dhlamb: Sure, but _what is the resource against which you are working_?
<dhlamb>ajs6f: a simple container
^simple as in brand new13:41
<ajs6f>dhlamb: So you did a POST, and got back an URI, and went from there?
<dhlamb>ajs6f: yes
<ajs6f>dhlamb: oaky.
dhlamb: Can you humor me? Do another POST, then do your "first query" and then do a GET and show us what you get back?13:45
<dhlamb>ajs6f: sure
I mean, dhlamb++.
<acoburn>dhlamb: silly question — when you use "dc" are use using dc or dcterms?13:46
<whikloj>ajs6f: I tried this and found the same thing
<ajs6f>whikloj: I believe that something is wrong. I'm trying to figure out what.
<ajs6f>whikloj: Don't get happy. It may mean a ticket for you.
<whikloj>expansion (tickets--)++13:48
<acoburn>dhlamb: I'm getting the same behavior that you describe13:50
<ajs6f>whikloj; Is tickets a value (in which case you said tickets==) or a reference to a value type (https://en.wikipedia.org/wiki/Value_type) in which case you called some kind of "static" method on a reference through "tickets—".13:51
<dhlamb>ajs6f: https://gist.github.com/daniel-dgi/31df45b5c8ac69423b5e
<ajs6f>Very important to get these things straight.13:52
<dhlamb>ajs6f: there's my results from the first query
acoburn: yeah, that's a drupal-ism i'll have to mitigate later. drupal uses the terms namespace with the 'dc' prefix and not 'dcterms'
acoburn: it's hardcoded in drupal core. literally. can't change it without hacking core code.13:53
<ajs6f>dhlamb: Okay, cool. So you got a single title, which is what we want. Now when you run the second query once, you get two titles, "things" and "stuff"?13:55
<dhlamb>ajs6f: yes
<ajs6f>dhlamb: Again, if you would humor me, would you run the second query but only the DELETE part, once? Don't do any insert, just the DELETE, and then a GET, and see what comes out?13:56
<dhlamb>ajs6f: np13:57
<ajs6f>dhlamb: Thanks!
(This is so funny! I have no idea what I'm doing. I'm not even the real Adam Soroka. I'm just some jerk who sat down in his office while he went to the restroom. I'm just going to see what I can get this dhlamb guy to do!)13:58
<jonroby>ajs6f - inside voice != outside voice
<ajs6f>dhlamb: Okay, would you help me out? I need you to go out into your office corridor and do the hoeky-pokey for a bit, and then do a GET on the resource, and then tell me what you get.
<dhlamb>ajs6f: do the delete with or without the author bit in the clause?14:00
ajs6f: i mean, is my right leg in or out?14:01
<ajs6f>dhlamb: Shouldn't matter to me: I want to see if there's an issue with the DELETE and the INSERT of props under the same name in the same operation. We've had problems with weird stuff with multivalued props, and that's what I'm sniffing at.
dhlamb: You're waving your right leg in the air like you just don't care.14:02
<dhlamb>ajs6f: kk
ajs6f: i had no idea i was so flexible. thought i lost that once i hit 30.
<ajs6f>dhlamb: You never lose the ability to just not care, if you keep exercising that muscle.14:03
<dhlamb>ajs6f: k, getting sucked into our comitters call, but i'll sneak another gist to you in just a sec
<ajs6f>dhlamb: NP. Thanks for exploring this crater.14:04
<dhlamb>ajs6f: https://gist.github.com/daniel-dgi/644f8b4ba9b20ca9babb14:07
ajs6f: looks like the delete went fine
<whikloj>ajs6f: I did the delete here: https://gist.github.com/whikloj/de4aeb73ec666be1527c14:08
<ajs6f>whikloj: You had two DELETE triples?14:09
dhlamb: The fact that the DELETE worked by itself is interesting. But the DELETE + the INSERT leaves two titles, right? But then you run it again and it leaves just the one, right?14:10
<whikloj>ajs6f: right
<ajs6f>whikloj: Why? You only had one title, right?
<whikloj>ajs6f: why 2 deletes?14:11
<ajs6f>whikloj: Right, right not just one <> dc:title ?someVar
But two?
* acoburn leaves
<whikloj>ajs6f: dhlamb's original was DELETE WHERE { <> dc:title ?o0 . <> dc:author ?o1 . }
ajs6f: one for title and one for author14:12
<ajs6f>whikloj: Right?
<whikloj>ajs6f: I think because the second RDF adds an author and a new title, dhlamb wants to remove the prior one. I just left that in the statement for now.14:13
<ajs6f>whikloj: What I'm saying is that dhlamb just did one <> dc:title ?var removal, but you seem to have done two. Right? That's not testing the same thing… I don't know that it really matters. I'm looking hard at the fact that it worked correctly by itself, but that an accompanying INSERT made it go wonky.14:15
I don't know in what order Jena does DELETEs and INSERTs or even if that order is defined.14:16
* github-ff joins
[ontology] yinlinchen opened pull request #28: Remove the fedora:Blanknode class from the Fedora ontology (master...fcrepo-1394) http://git.io/vkCgv
* github-ff leaves
<ajs6f>It shouldn't matter, _IF_ our handling of single- and multi-valued props is kosher.14:17
* travis-ci joins14:24
robyj/fcrepo-camel-toolbox#2 (fcrepo-1556 - b6feaf8 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo-camel-toolbox/compare/534b259d3e3c...b6feaf865e8c
Build details : http://travis-ci.org/robyj/fcrepo-camel-toolbox/builds/64460752
* travis-ci leaves
* github-ff joins
[fcrepo-camel-toolbox] robyj opened pull request #29: Fcrepo 1556 (master...fcrepo-1556) http://git.io/vkCw4
* github-ff leaves
<ajs6f>dhlamb: I think you got a bug by the tail here. Can you file a ticket? Do you want to write an iT, or would you rather I do? You've done the hard work of building the test scenario and workflow already.14:32
* scossu leaves14:35
* scossu joins14:41
<dhlamb>ajs6f: ok, out of my meeting. let me run through the steps a few more times (with the extra delete variable and such) to get it precisely worked through and i'll file the ticket.15:05
dhlamb: If we can't do property persistence, then we should turn it our badges and lay down our guns.15:06
* github-ff joins15:12
[fcrepo-webapp-plus] robyj opened pull request #12: Fcrepo 1556 (master...fcrepo-1556) http://git.io/vkChs
* github-ff leaves
* travis-ci joins15:15
fcrepo4-labs/fcrepo-audit#30 (fcrepo-1556.test - 34ee563 : Andrew Woods): The build failed.
Change view : https://github.com/fcrepo4-labs/fcrepo-audit/commit/34ee5631f7b3
Build details : http://travis-ci.org/fcrepo4-labs/fcrepo-audit/builds/64451706
* travis-ci leaves
<dhlamb>ajs6f: ok, pinned it down. looks like it short-circuits when you issue a DELETE WHERE with a property that doesn't exist yet. so if i have a dc:title, and then try to delete dc:title and dc:author when author doesn't exist yet, title does not get deleted15:22
ajs6f: is this still a bug, or expected behavior?
ajs6f: in other words, stop trying to delete things that don't exist, danny15:23
<ajs6f>dhlamb: No way that what you've been describing from the beginning is expected behavior, by the simple logic that you know more about Fedora than most people in the world and you did not expect it.
dhlamb: No, that's a buggy, awful response to what you were trying to do.15:24
<dhlamb>ajs6f: ok, making a ticket
ajs6f: good. makes my life a little easier. i've got a lot of situations where things may or may not exist and it's best to nuke from space to make sure i line up rdf between things.15:25
<ajs6f>dhlamb: Yes, rightly. And thanks for getting to the exact conditions that are responsbile. That will make the job of fixing it way, way, easier.
dhlamb: Deleting a non-existant property surely should not affect neighboring properties.
In the absence of inference, properties are orthogonal.
<dhlamb>ajs6f: well said15:26
to the jiratorium!
* travis-ci joins15:29
robyj/fcrepo4-oaiprovider#1 (fcrepo-1556 - 110725b : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo4-oaiprovider/compare/3049ce1967cc...110725bac81c
Build details : http://travis-ci.org/robyj/fcrepo4-oaiprovider/builds/64468369
* travis-ci leaves
* github-ff joins15:32
[fcrepo-audit] awoods deleted fcrepo-1556.test at 34ee563: http://git.io/vkWIK
* github-ff leaves
* github-ff joins15:44
[fcrepo4-oaiprovider] robyj opened pull request #18: Fcrepo 1556 (master...fcrepo-1556) http://git.io/vkWYs
* github-ff leaves
<dhlamb>thought a bot would spam the chan with it, but since not...15:57
* dhlamb leaves16:08
* travis-ci joins16:09
robyj/fcrepo4-swordserver#1 (fcrepo-1556 - 44f0e70 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo4-swordserver/commit/44f0e70a8fa6
Build details : http://travis-ci.org/robyj/fcrepo4-swordserver/builds/64474091
* travis-ci leaves
* github-ff joins16:12
[fcrepo4-swordserver] robyj opened pull request #5: - added javadoc plugin version 2.10.3 to pom.xml (master...fcrepo-1556) http://git.io/vkW82
* github-ff leaves
* jgpawletko leaves16:24
* jgpawletko joins
* dwilcox leaves16:33
* travis-ci joins16:36
robyj/fcrepo4-release-tests#1 (fcrepo-1556 - eb890a9 : robyj): The build passed.
Change view : https://github.com/robyj/fcrepo4-release-tests/commit/eb890a94fd5a
Build details : http://travis-ci.org/robyj/fcrepo4-release-tests/builds/64478011
* travis-ci leaves
* github-ff joins16:37
[fcrepo4-release-tests] robyj opened pull request #11: - added javadoc plugin version 2.10.3 to pom.xml (master...fcrepo-1556) http://git.io/vkWaV
* github-ff leaves
* osmandin leaves16:46
* ajs6f leaves16:48
* awead leaves16:49
* scossu leaves16:55
* scossu joins17:09
* travis-ci joins17:12
robyj/migration-utils#1 (fcrepo-1556 - 6061322 : robyj): The build has errored.
Change view : https://github.com/robyj/migration-utils/commit/6061322bb39d
Build details : http://travis-ci.org/robyj/migration-utils/builds/64483190
* travis-ci leaves
* travis-ci joins17:21
robyj/migration-utils#2 (fcrepo-1556 - 19e0a5b : robyj): The build passed.
Change view : https://github.com/robyj/migration-utils/compare/6061322bb39d...19e0a5b1c638
Build details : http://travis-ci.org/robyj/migration-utils/builds/64484934
* travis-ci leaves
* github-ff joins17:23
[migration-utils] robyj opened pull request #21: Fcrepo 1556 (master...fcrepo-1556) http://git.io/vkW58
* github-ff leaves
* awead joins17:24
* jgpawletko leaves17:35
* jonroby leaves17:36
* whikloj leaves17:51
* jgpawletko joins17:56
* jgpawletko leaves17:59
* dwilcox joins18:05
* awoods leaves18:12
* awoods joins18:24
* awoods leaves18:42
* ksclarke leaves18:46
* escowles leaves18:47
* awoods joins18:59
* awead leaves19:08
* scossu leaves19:11
* dwilcox_ joins19:41
* dwilcox leaves19:43
* dwilcox_ leaves20:01
* ksclarke joins20:20
* ksclarke leaves20:25
* ksclarke joins20:39
* awead joins20:45
* dhlamb joins21:07
* dhlamb leaves23:59
* mikeAtUVa leaves00:13
* mikeAtUVa joins00:17
* mikeAtUVa leaves00:27
* mikeAtUVa joins
* ksclarke leaves00:31

Generated by Sualtam