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

Using timezone: Eastern Standard Time
* barmintor leaves01:37
* ksclarke leaves02:14
* dshalvi leaves02:22
* awead_ joins04:23
* awead leaves04:24
* dhlamb joins06:37
* dhlamb leaves06:45
* dwilcox joins07:39
* barmintor joins08:04
* barmintor leaves
* jcoyne joins08:25
* dhlamb joins08:45
* jcoyne leaves08:46
* Nianli joins08:54
* whikloj joins08:57
* acoburn joins09:00
* Yinlin joins09:14
<Nianli>dwilcox: yesterday, I finished the final test on FCREPO-1371, and started the work on FCREPO-1376, today, will try to finish FCREPO-1376, and start to look at other ticket. Blocker: I have some local work need to be done asap today, but I will make up some time tonight.09:28
<Yinlin>Why does FCREPO-1366 just change 3030 port in https://github.com/fcrepo4/fcrepo-message-consumer/commit/6d6a9ef11330f246e00b69e0e1020c2b1b8b693409:51
Does Fuseki need only use 3030 port?
Just curious, what do you want to solve in this ticket?09:52
awoods: forget to add you name
<awoods>on a call09:53
<Yinlin>awoods: np
<whikloj>general question: I have a JCRNode and an RDFNode, what is the easiest way to check if they are the same item? Assuming I need to convert one.09:58
* ksclarke joins10:04
* dwilcox leaves10:26
* barmintor_ joins10:28
* dwilcox joins10:34
* jcoyne joins10:40
<f4jenkins>Project fcrepo4-T2 build #166: UNSTABLE in 7 min 4 sec: http://jenkins.fcrepo.org/job/fcrepo4-T2/166/10:55
* awoods: Enable LDP indirect containers to update referenced container
* awoods: Add unit test for IndirectContainer with reference property
<whikloj>awoods/dwilcox: Yesterday finished (hopefully) FCREPO-1377, working on FCREPO-1362 but need some clarification on comparing JCRNodes vs RDFNodes.11:00
awoods/dwilcox: See https://github.com/whikloj/fcrepo4/blob/FCREPO-1362/fcrepo-kernel-impl/src/main/java/org/fcrepo/kernel/impl/FedoraResourceImpl.java#L275
* mikeAtUVa joins11:01
* ajs6f_ joins11:14
barmintor: I don't want to freak you out, but I just disagreed with you on the mailing list.
whikloj: If at all possible, you want to deal with RDF and Fedora model types (like FedoraResource), not JCR types.11:16
whikloj: Those canbe converted with an IdentityConverter, I think.
type?whikloj: How did you end up with a naked JCR
type?
<barmintor>ajs6f_: I think what you say is illustrating my point, actually.11:17
* terrellt leaves11:19
<ajs6f_>barmintor: Well, like I did say, I agree with you as far as implementation goes. I just don't think that's the whole story.11:20
* terrellt joins11:21
<barmintor>I think it is literally the whole story.11:23
then again, I think Fedora has been on a long arc towards RDF since v2, but this is not a universally shared opinion11:32
<jcoyne>ajs6f_, awoods, barmintor: speaking of, I've found that I need to run an additional triple store (marmotta) to cache remote (id.loc.gov) graph fragments. It seems awkward. It would be swell if Fedora had a solution to that.11:35
<barmintor>jcoyne: I’m not sure what you mean. What do you want Fedora to do?11:36
<jcoyne>barmintor: solve all my problems.
<barmintor>jcoyne: open a ticket.
jcoyne: that’s an epic, make sure you’re the reporter. Assign it to ajs6f_11:37
<jcoyne>barmintor: :)
<ajs6f_>barmintor: An arc towards RDF, yes. An arc towards the abandonment of modeling as a practice, that's not so clear to me.11:39
<jcoyne>barmintor: Honestly, It seems like other people must have this problem too. I don't know if the solution is writing more code in Fedora. I guess I'm looking for a recipe of "how do we solve this problem"
<barmintor>ajs6f_: The Object/DS distinction has nothing to do with the kind of modeling you’re talking about, though
<ajs6f_>jcoyne: Fedora is a repository, not a web cache. I don't mean that in a snarky way, I really mean that using Fedora to cache linked data is like using a jet fighter to go get your groceries. Really expensive.11:40
barmintor: Obviously, I disagree. It is exactly at the heart of that modeling.
* Yinlin leaves
<jcoyne>ajs6f_: Fair enough. But are you having the problem I have? How are you solving it?11:41
<ajs6f_>barmintor: I'm talking about modeling in the Fedora sense, not modeling in some object-oriented sense or in some ontologically-driven sense. Modeling for durability.
<barmintor>I know. I’m not communicating to you very well.
<ajs6f_>jcoyne: Sure, and I'm solving it with the Apache Stanbol EntityHub, but that's just my choice. I think it's quite good tho-- worth a look.
barmintor: Let's leave it be. as long as Aaron B. gets where he's trying to go, we can leave it for over a beer, eh?11:42
<mikeAtUVa>awoods, barmintor, would it make sense to have a quick chat to hash out the ISPN cache idea so we're all on the same page?
<jcoyne>ajs6f_: ah, cbeer mentioned that the other day.
<ajs6f_>jcoyne: If cbeer eats it, you know it's tasty and good for you.11:43
<awoods>mikeAtUVa: sounds good
<jcoyne>ajs6f_: it sounds like he's still making up his mind about it... so It may be poison.
<ajs6f_>jcoyne: Let me know if he gets sick. If he does, I'll turn off my instances. Seriously, it's one way to go, but Marmotta is pretty good too, and there are other ways. Just a simple Jena instance might be what you need.11:44
jcoyne: All I'm saying is that putting everything in Fedora might be a false economy.11:45
jcoyne: Remember that Fedora has no real internal querying-- you need to index it anyway. Why not drive that stuff right into the index, instead of through your repository?11:46
<jcoyne>ajs6f_: Yeah, I'm pretty convinced of that, I'm just saying that we have another problem around caching linked data.
ajs6f_: right, I considered writing it right to solr too.
<ajs6f_>jcyone: Sure, that's a real problem, a big one that we share with a big commuinity. I'm not sure it's a good diea for Fedora to invent its own solutions to that problem, but we should certainly have some community knowledge around it.11:47
jcoyne: FWIW EntityHub can be backed directly by Solr.
jcoyne: That's how I run it.
<whikloj>ajs6f_: sorry for the delay, the JCRNode is the node being deleted. It is actually a type "Node" which is a JCR "item".11:55
ajs6f_: This one https://github.com/fcrepo4/fcrepo4/blob/master/fcrepo-kernel-impl/src/main/java/org/fcrepo/kernel/impl/FedoraResourceImpl.java#L91
<barmintor>ajs6f_: I have a sincere question re: FCR3 CMA and services, do you have a minute for a PM?11:58
* awoods leaves
<acoburn>jcoyne ajs6f_: I also use a stanbol entityhub for precisely this purpose. In my case, for LoC named authorities and the geonames dataset12:02
jcoyne ajs6f_: also with a solr backend
* awoods joins12:11
* jcoyne leaves12:12
<ajs6f_>barmintor: Sure.12:14
whikloj: I understand that. I'm asking why you have that in hand and not a Fedora model class wrapping it. Where did you get the JCR node?12:18
* jcoyne joins
<awoods>ajs6f_: whikloj is working in kernel-impl: https://github.com/whikloj/fcrepo4/blob/FCREPO-1362/fcrepo-kernel-impl/src/main/java/org/fcrepo/kernel/impl/FedoraResourceImpl.java#L27512:19
<whikloj>ajs6f_: In the delete method that is the "node" used to get the weak references
ajs6f_: or what awoods said12:20
<ajs6f_>whikloj: and what are you trying to convert to an RDF node?12:23
<whikloj>ajs6f_: the weak reference property value12:24
<ajs6f_>whikloj: If it's a Value you are trying to convert, why not use propCvt?12:25
<whikloj>ajs6f_: I use that to convert the property value to an RDFNode, but I want to compare that to the object we are deleting, which seems to be a JCRNode.12:26
ajs6f_: big picture. This object has a multivalue weak reference pointing at it. I need to remove it from the property values without removing the whole property12:28
in case I am going at it all wrong
<ajs6f_>whikloj: Wait, pointing _at_ it? Meaning this object is the thing you are deleting?12:30
<awoods>ajs6f_: whikloj is working: https://jira.duraspace.org/browse/FCREPO-136212:31
<whikloj>ajs6f_: yes, that is my understanding of it
<ajs6f_>whikloj: Then why are you starting by retrieving the JCR node under 'this'? why not just start with 'this'?12:33
I.e. go to the other node's property and delete the value in it of this.getNode()?
I don't think I understand what's going on here12:34
<whikloj>ajs6f_: No I think you do, I think I don't have full grasp of how the property references both nodes12:35
<ajs6f_>The value of the multivalued weak reference prperty is a bunch of nodes, right?
and you are deleteing one of them
(One of the nodes, that is)
<whikloj>yes, Value[] of nodes
<ajs6f_>It happens to be the one that underlies the instance of the class that you are actually writing in, right?
<whikloj>right12:36
<ajs6f_>So go to the property, get the Value[], remove the Value that matches this.getNode(), and move on, no?
<whikloj>right, but how do you do that comparison of Value == Node12:38
identifier?
or actual object?
<ajs6f_>The Value is a WEAKREFERENCE, right?12:39
<whikloj>the property is WEAKREFERENCE, the Value is a node
or rather the values are...something that can be converted to a RDFNode12:40
<ajs6f_>http://www.day.com/maven/jsr170/javadocs/jcr-2.0/index.html?javax/jcr/PropertyType.html12:41
Says that the value in the Value is an identifier for the node referenced, so use Session.getNodeByIdentifier(), why not?12:42
<whikloj>ajs6f_: because I didn't know I could. So thanks, I'll try that :)
<ajs6f_>whikloj: If your machine exp[lodes in flames and dies, I never even heard of you.12:44
<whikloj>ajs6f_: I'm writing your name on the wall in blood right now.
<barmintor>wow that’s dark
<whikloj>and I haven't even tried the code yet12:45
<ajs6f_>Fedora 4: Put a little Lovecraft in your life.12:46
* jcoyne leaves12:51
<ruebot>barmintor: finished testing fcrepo-1393 -- let me know if this is sane https://gist.github.com/ruebot/b3a5a2ed391f4991f1ba
* jcoyne joins12:56
* github-ff joins
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/pvkr
fcrepo4/master bd5b35a Yinlin Chen: Enhance RDF parsing error message to client....
* github-ff leaves
* github-ff joins12:57
[fcrepo4] awoods closed pull request #742: Fcrepo 1382 (master...fcrepo-1382) http://git.io/xdHs
* github-ff leaves
* acoburn leaves13:27
* acoburn joins13:32
<barmintor>Is the audit conversation going to include a proposal that describes handling it as an indirect container, or some other LDP-centric approach?13:41
* dwilcox leaves13:42
* acoburn leaves13:49
* Yinlin joins13:57
<ajs6f_>barmintor: That sounds good to me, but it's @dshalvi who has been looking at the API, I think. Perhaps you can comment on his page?14:01
<barmintor>Hmm, I felt like the list was the more appropriate place to bring it up for transparency’s sake14:04
* awoods leaves14:17
<ajs6f_>barmintor: agreed, I meanbt in addtion, just so that people looking at that page get a sense of discussion.14:20
<barmintor>ajs6f_++
* barmintor noted it
I think it would also be wise to ask someone from LOC whether http://id.loc.gov/ontologies/premis.html#hasRelatedObject isn’t the right membershipRelation to point from an LDPC of PREMIS events to the object they describe14:21
<ajs6f_>barmintor: +114:23
ruebot: Is that in your bag, or is that further than the scope of your comparison?
* ruebot looks14:25
* ruebot agrees barmintor14:27
* ruebot mumbles horrible things about PREMIS Ontology
(We just use PREMIS XML in Islandora, lot more sane. No proliferation of classes.)
* barmintor cares more about limiting idiosyncratic API developments than which ontology to use14:28
http://id.loc.gov/ontologies/premis.html#hasRelatedStatuteInformation now we’re talking14:29
<ajs6f_>There _are_ standards worse than making your own, but not many.
* awoods joins14:30
<ruebot>It'd be helpful if the PIG mailing list wasn't so dead...14:31
<barmintor>There’s still a hi-def PIG14:33
https://wiki.duraspace.org/display/hydra/Hydra+Digital+Preservation+Interest+Group
* jgpawletko joins14:38
* acoburn joins14:55
<ruebot>PIG FIGHT!15:08
https://github.com/Islandora/Islandora-Preservation-Interest-Group
<Yinlin>awoods: free?15:17
* dhlamb leaves15:18
* acoburn leaves15:23
* Yinlin leaves15:31
* github-ff joins16:38
[fcrepo4] whikloj opened pull request #744: Surgical multi-relationship removal - Addresses FCREPO-1362 (master...FCREPO-1362) http://git.io/pJJE
* github-ff leaves
* mikeAtUVa leaves
<whikloj>awoods/dwilcox: If you have a preference for my next ticket let it be known now.16:40
<ajs6f_>Have a gopod weekend all...16:43
or a good weekend, even
* ajs6f_ leaves
* scossu joins16:46
* awead leaves16:52
<barmintor>ruebot: iPIG is a Mac thing, right?17:05
<ruebot>barmintor: I got 99 problems and a Mac ain't one :-P
* jcoyne leaves17:08
* whikloj leaves17:54
* jcoyne joins18:00
* dhlamb joins18:40
* ksclarke leaves
* jcoyne leaves18:42
* dhlamb leaves18:47
* scossu leaves18:58
* dhlamb joins19:33
* whikloj joins19:45
* whikloj leaves20:05
* dhlamb leaves20:08
* whikloj joins20:28
* ksclarke joins20:33
* jgpawletko leaves20:51
* jcoyne joins20:52
* whikloj leaves20:53
* jcoyne leaves20:55
* whikloj joins21:24
* jcoyne joins21:55
* jgpawletko joins22:25
* whikloj leaves22:28
* jgpawletko leaves22:48
* jcoyne leaves23:12
* jcoyne joins
* jcoyne leaves23:34
* beermintor_ joins00:07
* bljenkins leaves00:09
* bljenkins joins00:10
* beermintor leaves

Generated by Sualtam