Log of the #duraspace-ff channel on chat.freenode.net

Using timezone: Eastern Standard Time
* martinjd leaves02:41
* martinjd joins05:31
* martinjd leaves06:54
* jcoyne joins07:11
* jcoyne leaves07:13
* scossu joins07:37
* dwilcox joins07:55
* jcoyne joins08:16
* scossu leaves08:19
* scossu joins08:20
* scossu leaves08:38
* scossu joins08:47
* scossu leaves
* mikeAtUVa leaves09:06
* mikeAtUVa joins09:10
* ermadmix joins
* ksclarke joins09:21
* gregjansen joins09:35
* github-ff joins09:39
[fcrepo4] cbeer created html-ui (+1 new commit): http://git.io/o0WqkQ
fcrepo4/html-ui 84cfe19 Chris Beer: Update HTML form for creating new content to use PUT/POST as appropriate.
* github-ff leaves
* github-ff joins
[fcrepo4] cbeer opened pull request #333: Update HTML form for creating new content to use PUT/POST as appropriate... (master...html-ui) http://git.io/DRLCRg
* github-ff leaves
<pivotal-bot_>Chris Beer added comment: "https://github.com/futures/fcrepo4/pull/333" https://www.pivotaltracker.com/story/show/70492104
Chris Beer started "Fix UI from POST updates" https://www.pivotaltracker.com/story/show/7049210409:41
Chris Beer finished "Fix UI from POST updates" https://www.pivotaltracker.com/story/show/70492104
Chris Beer edited "Upgrade to Modeshape 3.7.3" https://www.pivotaltracker.com/story/show/7040187209:43
* travis-ci joins09:51
[travis-ci] futures/fcrepo4#1867 (html-ui - 84cfe19 : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/commit/84cfe19a699b
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/24182621
* travis-ci leaves
* scossu joins10:06
* scossu1 joins10:08
* scossu leaves10:11
<awoods>cbeer: I am getting FedoraHtmlResponsesIT test failures in PR-33310:12
<pivotal-bot_>Andrew Woods added comment: "Failed tests:
FedoraHtmlResponsesIT.testCreateNewNodeWithProvidedId:89->createAndVerifyObjectWithIdFromR..." https://www.pivotaltracker.com/story/show/70492104
Andrew Woods rejected "Fix UI from POST updates" https://www.pivotaltracker.com/story/show/70492104
Andrew Woods accepted "Refactor modules out of main code base" https://www.pivotaltracker.com/story/show/70092316
Andrew Woods added "Migrate fcrepo-generator-dc functionality" https://www.pivotaltracker.com/story/show/7051037610:20
* scossu1 leaves10:21
* scossu joins
* edInCo joins10:37
* dwilcox leaves10:38
<awoods>do you have something to share, ksclarke?10:42
<pivotal-bot_>Andrew Woods started "Transition jetty integration tests to grizzly" https://www.pivotaltracker.com/story/show/7010043810:43
<ksclarke>just my complete lack of understanding of rdf :-(
I'm curious about that ticket though ^ what's prompting the move from jetty to grizzly?
<awoods>ksclarke: our build is too slow10:44
<ksclarke>huh, so grizzly is faster then I take it? interesting
<awoods>ksclarke: yes.
<jcoyne>ksclarke: have you looked at active-tripples10:54
<ksclarke>jcoynes: I haven't10:55
* ksclarke goes to look
:-) I'm not really a ruby person either though
<jcoyne>s'pretty cool
well, you don't have to get very invested in ruby.
ksclarke: Tom Johnson & Trey Terrell, they know a lot about RDF10:56
It basically gives you a way to have a RDF ORM10:57
* dwilcox joins
<awoods>ksclarke: it may be useful to start pulling together some links to add to a "resources" wiki page.10:58
* escowles joins10:59
* osmandin joins11:00
* gregjansen leaves11:05
* gregjansen joins11:20
<escowles>gregjansen: https://gist.github.com/mjgiarlo/709616011:25
* osmandin leaves11:59
* gregjansen leaves12:04
* gregjansen joins12:05
thanks escowles
escowles: not sure I should jump into that thread, but the machine readable "rightsHolder" or rights statements have always sounded more like roles to me.. More of a software term than a intellectual property term, but it seems like these are expressing different things to me..12:08
rights: broad statements about who (public mostly, perhaps educators or not-profit people) can do what (redistribute, make copies, etc..)12:10
roles: classes of people that can take certain actions in a system, maps loosely to the "who" above. (Some rights may be enforceable via system roles)
awoods, escowles: might also be worth putting giarlo in touch with Martin, who was talking extensively about machine readable rights RDF last week.12:13
<awoods>gregjansen: on a call12:18
<gregjansen>awoods: off to lunch, ttys12:19
<scossu>Minutes are online: https://wiki.duraspace.org/display/FF/2014-05-01+-+Fedora+Committer+Meeting12:54
Feel free to add or edit if I missed somehting.
<awoods>thanks, scossu12:55
* github-ff joins13:01
[fcrepo4] cbeer force-pushed html-ui from 84cfe19 to fefb1ce: http://git.io/gHQ14w
fcrepo4/html-ui fefb1ce Chris Beer: Update HTML form for creating new content to use PUT/POST as appropriate.
* github-ff leaves
* ermadmix leaves13:04
* travis-ci joins13:15
[travis-ci] futures/fcrepo4#1870 (html-ui - fefb1ce : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/84cfe19a699b...fefb1cee62ed
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/24198107
* travis-ci leaves
* ermadmix joins13:26
* ermadmix leaves13:58
* ermadmix joins14:02
* dwilcox leaves14:04
* dwilcox joins14:17
<pivotal-bot_>Andrew Woods delivered "Fix UI from POST updates" https://www.pivotaltracker.com/story/show/7049210414:24
* github-ff joins
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/PbIe6A
fcrepo4/master 21d0b88 Andrew Woods: Merge pull request #333 from futures/html-ui...
* github-ff leaves
* travis-ci joins14:40
[travis-ci] futures/fcrepo4#1872 (master - 21d0b88 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/5b4d2e14ca46...21d0b8888cb7
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/24204375
* travis-ci leaves
<awoods>jcoyne: I just did a comb-through of the build, and am only seeing jetty being started in the build process in the fcrepo-webapp project. I suspect the other start-ups you were seeing were actually Grizzly.14:46
<jcoyne>awoods: I was just repeating what I heard barmintor or escowles say14:48
<awoods>jcoyne: ok. In any case, it would be nice to have a quicker build.14:49
<jcoyne>and not run out of permgen14:50
<escowles>ok, maybe it's just grizzly getting started multiple times14:52
<awoods>jcoyne: that would also be a treat, for those that would rather not tweak JAVA_OPTS settings.
* escowles leaves15:23
* nikhiltri joins16:16
<pivotal-bot_>Andrew Woods deleted "Transition jetty integration tests to grizzly" https://www.pivotaltracker.com/story/show/7010043816:18
Andrew Woods accepted "Fix UI from POST updates" https://www.pivotaltracker.com/story/show/7049210416:19
<awoods>scossu: in your CND: https://github.com/scossu/artic_fedora_cnd/blob/master/classes/cnd/aic-assets.cnd16:31
* dwilcox leaves
<awoods>scossu: in your definition of the P70_documents property, for example...
- cidoc:P70_documents (reference) mul < aic:nonasset
<awoods>scossu: my impression of the "<" syntax on properties was a statement fof "constraint"
scossu: what did you have in mind for "< aic:nonasset"?16:33
<scossu>awoods: that's a constraint on the node type that can be assigned to that relationship.
a little bit like rdfs:range.
<awoods>scossu: what is the constraint, in this case?16:34
scossu: I am not seeing aic:nonasset defined anywhere.16:35
<scossu>awoods: https://github.com/scossu/artic_fedora_cnd/blob/master/classes/cnd/aic-resources.cnd#L58
awoods: it's a special constraint that applies to references and weakreferences only.16:36
<awoods>scossu: got it. thanks.16:37
ksclarke: do you have some source line numbers relating to this ticket?16:38
<ksclarke>awoods: here was where I was looking - https://gist.github.com/ksclarke/73d85496019377669d88
they're small classes
those two first toTriples classes looked like the distinction scossu was making, so was investigated getting another triple out of them for rdf:type16:40
investigating, that is
<awoods>ksclarke: those seem like the places to start.16:41
<ksclarke>including another triple out in the RdfStream? or was there a better way?16:42
<awoods>ksclarke: I would like to see if we can include details from the node/property definitions themselves, as opposed to throwing in a fabricated triple.16:44
ksclarke: It looks like this is where the errant "Property" triple is inserted: https://github.com/futures/fcrepo4/blob/master/fcrepo-kernel/src/main/java/org/fcrepo/kernel/rdf/impl/mappings/ItemDefinitionToTriples.java#L79
ksclarke: google-hangout?16:46
* mikeAtUVa leaves16:57
* gregjansen leaves17:03
<awoods>ksclarke: I think your suspicion is right...17:28
ksclarke: <http://www.jcp.org/jcr/nt/1.0nodeType> seems ideal.
<ksclarke>okay, I'll work with that and if cbeer or esc say differently I can swap out17:29
<awoods>ksclarke: sounds good.17:30
cbeer: do you have a take on this...
<cbeer>awoods: is there a summary somewhere?17:31
<awoods>cbeer: Let me lay it out...
<cbeer>email to the list, maybe?
<awoods>cbeer: Here we define the triple for items: https://github.com/futures/fcrepo4/blob/master/fcrepo-kernel/src/main/java/org/fcrepo/kernel/rdf/impl/mappings/ItemDefinitionToTriples.java#L7917:32
cbeer: There is an incorrect assumption that this method will only be executed for "Properties"
cbeer: This method is also executed for "Nodes"
<awoods>cbeer: Child nodes, in the CND sense17:33
<cbeer>aren't child nodes properties?
<awoods>are you suggesting that there is no concept of a "node" different than the concept of a "property"?17:34
<cbeer>i'm not clear what distinction you're trying to make
<awoods>A node-type can define a set of properties, and a set of child nodes.17:35
cbeer: the distinction is between those two things.
<cbeer>awoods: isn't the child node definition just a property that happens to point at a node?17:36
or are we talking about some other type of thing?
<awoods>cbeer: https://docs.jboss.org/author/display/MODE/Defining+custom+node+types17:37
cbeer: There you see "Property definitions" and "Child node definitions"
cbeer: There is a separate syntax for "properties" (-) and for "child nodes" (+)17:38
cbeer: The types for properties and nodes are also different
cbeer: Personally, I think there is a clear distinction between properties and nodes (or child nodes in this case). Are you seeing it differently?17:40
<cbeer>i see the distinction in the CND. i don't see the distinction in the JCR API, and I don't see the distinction we'd want to offer downstream in an RDF vocabulary17:41
<awoods>cbeer: You would prefer not to describe nodes as: <http://www.jcp.org/jcr/nt/1.0nodeType> ?17:42
<cbeer>awoods: "describe nodes"?
sorry, i feel like i'm missing a whole lot of context17:43
<awoods>cbeer: https://www.pivotaltracker.com/story/show/69346904
<pivotal-bot_>feature: Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint (started) / owner: Kevin Clarke
<ksclarke>cbeer: and the relevant part of his CND is pulled out here --> https://gist.githubusercontent.com/ksclarke/a46a240cc02615f987ee/raw/61cc989f222f5848512f8ed9e0af5936bec99b76/trying-to-grok.txt17:44
<cbeer>awoods: ok... i don't understand why a downstream consumer would need to be concerned about that, though?17:45
<awoods>cbeer: we expose the /fcr:nodetypes endpoint
cbeer: the descriptions in the response to that endpoint should include useful and accurate information17:46
<cbeer>ok. i guess i don't understand what's not useful or accurate about the triples in that ticket
<awoods>cbeer: interesting17:47
<cbeer>it seems like a reasonable reflection of what we're exposing over the API, isn't it?
or, wait.
sorry, trying to understand the CND17:48
awoods: yeah, i don't see a problem. as far as an RDF consumer is aware, those things are just rdf:properties, and the range tells you what you'll find there.17:49
<awoods>cbeer: There is an argument for using the combination of rdf#type and rdf#range.17:50
<ksclarke>cbeer: his response from that question yesterday was that there will be a variety of possible values for range and he wants to know what which group it falls into without iterating over many possible values
<awoods>"his" being scossu17:51
<cbeer>awoods: rdf#type? of <http://artic.edu/definitions/ontology/1.0#ds_master>?
ksclarke: so the problem is not having a way to resolve <http://www.jcp.org/jcr/nt/1.0file> into a JCR item, i guess?17:52
<awoods>cbeer: In this case, it seems like the rdf#type could be <http://www.jcp.org/jcr/nt/1.0nodeType>
cbeer: as opposed to <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property>17:53
<ksclarke>nt:file will be subclassed (probably the wrong word) into many different possibilities; yes, cbeer
<cbeer>awoods: that doesn't make sense to me. we're exposing <http://artic.edu/definitions/ontology/1.0#ds_master> as an rdf:Property.
<awoods>cbeer: what does not make sense?
<cbeer>unless you're suggesting we define<http://www.jcp.org/jcr/nt/1.0nodeType> as a subtype of rdf:Property?
but that seems much more difficult for naive rdf clients to parse17:54
<ksclarke>ping scossu
<awoods>cbeer: It seems like the fundamental difference here is between viewing child nodes as properties or nodes.17:55
cbeer: You view child nodes as properties, correct?
<cbeer>awoods: unless i'm totally misunderstanding how child nodes are actually represented in the RDF we're exposing, i don't see the difference between a property that is a string and a property that defines a child node.17:56
<awoods>cbeer: I can understand that perspective.17:57
<cbeer>awoods: that's good. i don't understand the one you're describing.17:58
<awoods>cbeer: hmm
cbeer: would you view the pivotal ticket as a non-issue?17:59
<pivotal-bot_>feature: Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint (started) / owner: Kevin Clarke
<cbeer>awoods: i don't understand the user story behind it, no.
<ksclarke>I think it's he wants to know the difference between string and child node (with lots of variations for different types of child nodes) without having to iterate through all the possibilities
<cbeer>ok. i'm not sure why you'd need to discover that difference, what good it does you to know that, and i don't think finding those things out is really that bad with a decent RDF client (assuming we're exposing good rdfs.)18:00
* ksclarke doesn't know the answer to that18:01
<awoods>I need to step away. We can pick it up tomorrow. I will be interested to hear escowles perspective as well.18:02
<ksclarke>thanks, awoods and cbeer, for your help
<cbeer>i certainly have one of those.18:03
<scossu>ksclarke: back from a meeting, catching up with the discussion...
<ksclarke>scossu: cbeer's last question might be most relevant: "ok. i'm not sure why you'd need to discover that difference, what good it does you to know that, and i don't think finding those things out is really that bad with a decent RDF client (assuming we're exposing good rdfs.)"18:04
* ermadmix leaves
<scossu>ksclarke, cbeer: an use case is when I want to know which child nodes a determined node type provides without making assumptions.18:05
"assumptions" being, as we discussed before, that a child node can also not be a nt:file. So just checking rdfs:range might not be enough.18:06
<ksclarke>makes sense to me but I'm not an experienced rdf user so can't speak to cbeer's third question/statement18:07
<cbeer>scossu: but all nodes should be nt:base, right?18:09
<scossu>I'm not sure if child node definitions can per se be properties; but in any case, if we consider a child node definition as a sub-property of a property definition, that should be a good enough distinction.18:10
cbeer: yes, but what does that imply?
The purpose is to be able to query schema information.18:12
Dumping the contents of /fcr:nodetype into a triplestore is very simple. That triplestore can be used to query the schema and all kinds of inferences can be made on them.18:13
<cbeer>scossu: if every child node definition's range is (ultimately) an nt:base, you should be able to discover what things are child nodes by following the RDFS up to the nt:base definition (also in fcr:nodetypes)
and if you're actually in a triplestore with inferencing, it ought to be trivial
<scossu>cbeer: uhm, I didn't think about that. I can experiment on the data that I have.18:14
<cbeer>?a <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property> ?b; ?b <rdfs:range> <nt:base>
* nikhiltri leaves18:22
<scossu>I can't see any of my nodes that have <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property> as a predicate.
<cbeer>scossu: sorry, missed a level.18:23
?a a <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property>; ?a <rdfs:range> <nt:base>18:24
(and, really, you'd want ?a <rdfs:domain> <the:node:i:care:About>18:25
* edInCo leaves
<scossu>Oh, that's right.18:27
cbeer: I think I have to build those relationship manually, because the only node that shows up with rdfs:range nt:base is fcrepo:content.18:28
Yup - the only rdfs:range value that I have for aic:ds_master, which is a child node, is nt:file. The supertypes of nt:file are not showing.18:31
Actually not, I can see the relationships. So the query would be18:37
?cn rdfs:range ?r .
?r rdfs:subClassOf nt:base .
I can see some system stuff, but I should be able to filter it out by specifying the domain.18:38
?cn rdfs:domain <my node type definition> .
?cn rdfs:range ?r .
?r rdfs:subClassOf nt:base .
This works great!18:45
<awoods>scossu/cbeer/ksclarke: That is good to hear.18:57
scossu: would you care to delete the ticket?18:58
<pivotal-bot_>feature: Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint (started) / owner: Kevin Clarke
<scossu>awoods: Shall I delete it or post the SPARQL solution and mark it as resolved?19:00
<awoods>scossu: post and resolve.19:01
<pivotal-bot_>Stefano Cossu added comment: "As per @cbeer 's suggestion this can actually be accomplished by dumping the contents of /fcr:nodetypes to..." https://www.pivotaltracker.com/story/show/6934690419:03
Stefano Cossu delivered "Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint" https://www.pivotaltracker.com/story/show/6934690419:04
Stefano Cossu finished "Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint" https://www.pivotaltracker.com/story/show/69346904
<scossu>Sorry for dragging you guys in this long discussion while the solution was there already...19:05
<pivotal-bot_>Andrew Woods estimated "Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint" as 2 points https://www.pivotaltracker.com/story/show/6934690419:06
Andrew Woods accepted "Distinguish between properties and child nodes in fcr:nodetypes REST API endpoint" https://www.pivotaltracker.com/story/show/69346904
<awoods>scossu: It was a team building experience.
<scossu>awoods, that's the best euphemism for "trolling" I have heard so far.19:10
* jcoyne leaves19:11
* jcoyne joins19:33
* awoods leaves19:51
* awoods joins19:54
* awoods leaves
* scossu leaves19:59
* ksclarke leaves20:19
* awoods joins20:50
* scossu joins22:44