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

Using timezone: Eastern Standard Time
* ksclarke leaves00:21
* kaarefc joins03:34
* kaarefc leaves04:18
* kaarefc joins04:55
* kaarefc leaves05:35
* kaarefc joins05:47
* jay joins08:23
* gregjansen leaves09:01
* escowles joins09:32
<pivotal-bot____>Yuqing Jiang finished "implemente add/get relationships for Tuque" https://www.pivotaltracker.com/story/show/5503868009:46
Yuqing Jiang added comment: "https://github.com/yqjiang/tuque/commit/ca12dd3453c8b74321988c20226e6dba41e7676c" https://www.pivotaltracker.com/story/show/5503868009:47
* ksclarke joins09:48
* ksclarke leaves09:51
<pivotal-bot____>Yuqing Jiang estimated "generate dc in tuque" as 3 points https://www.pivotaltracker.com/story/show/55038806
Yuqing Jiang estimated "implemente add/get relationships for Tuque" as 3 points https://www.pivotaltracker.com/story/show/55038680
* ksclarke joins09:52
<pivotal-bot____>A. "Cartroef" Soroka added comment: "I'm confused: http://fcrepo.org/ontology/ gives a 404. Is it necessary to go directly to a particul..." https://www.pivotaltracker.com/story/show/5509558009:55
* kaarefc leaves09:58
<pivotal-bot____>Andrew Woods added comment: " http://fedora.info/definitions/v4/repository09:59
http://fedora.in..." https://www.pivotaltracker.com/story/show/55095580
A. "Cartroef" Soroka added comment: "Okay. We should spread links liberally over the human-readable portions of the various websites, eh?" https://www.pivotaltracker.com/story/show/5509558010:01
* kaarefc joins10:22
* kaarefc leaves10:53
* nbanks joins10:54
* kaarefc joins11:06
* jongibson joins11:08
<pivotal-bot____>Andrew Woods edited "implement add/get relationships for Tuque" https://www.pivotaltracker.com/story/show/5503868011:14
* barmintor joins11:17
* kaarefc leaves11:19
<barmintor>awoods: the closest I've come to recs for housing in Austin are for 5 BR places just south of the River (that's more-or-less 20 blocks from the DLF facility)11:26
<awoods>barmintor: hmmm
<barmintor>awoods: Also had people recommend some common workspace that lets you rent private conference rooms, but that would require sep housing option11:27
<awoods>barmintor: Are these houses from air-bnb or from your friends?
<barmintor>I'll keep looking, thi
the houses are from… another site. but friends rec'd them
jongibson can attest to the difficulty of finding multi-room residences near campus, I'm jsure.11:29
<awoods>barmintor: As it stands right now, we just need a 5-bedroom place.
<barmintor>awoods: if we had, say, a 2BR and a 3BR place, but workspace at the UT library, wuld that work?11:30
<awoods>escowles: Could you briefly describe the issue(s) you were running into with the SPARQL query endpoint, and your next steps?
<barmintor>: I'l sendmyouthe links tonight
<awoods>barmintor: I suspect two places near one-another could work.11:31
<nbanks>Hey Andrew, could I get push rights to https://github.com/futures/tuque11:33
<awoods>nbanks: what is your git handle?11:36
nbanks: nevermind, got it11:37
nbanks: You should be good now.11:39
* github-ff joins11:42
[tuque] nigelgbanks deleted break_out_implementations at 9b62a87: http://git.io/2QJkCQ
* github-ff leaves
<nbanks>Can I also get admin access to change the base branch to fedora4? Or could you do that for me?11:43
<awoods>nbanks: You should be golden.11:48
<nbanks>awoods: I don't seem to have admin or push privledges anymore?12:05
<awoods>https://github.com/NigelBanks ?12:07
nbanks: https://github.com/nigelgbanks12:08
nbanks: try now
* jongibson leaves12:09
* jongibson joins12:21
<awoods>escowles: ^^^ Did you see the request above?12:25
<escowles>awoods: sorry -- no
<awoods>escowles: Could you briefly describe the issue(s) you were running into with the SPARQL query endpoint, and your next steps?12:26
<escowles>the problem i was having is that SPARQL update doesn't seem to be deleting blank node children
<awoods>escowles: Are you handling update as well as query?12:27
<escowles>i tried various permutations of different formats USING DELETE DATA, or doing a DELETE WHERE , etc.
<awoods>escowles: I thought you were trying to tackle SPARQL query.
<escowles>awoods: yes, my plan was to support updates and deletes
no -- i'm working on the indexer to sync updates based on update/delete events12:28
i thought we agreed that a separate triplestore could just be queried separately and there wasn't any rationale for providing a proxy
<awoods>escowles: yes12:29
hmm, simple SPARQL update/delete events are currently supported in the Fedora API
<escowles>awoods: the problem is when there are blank nodes in an object, I haven't been able to address them, so they get left behind12:30
i'm working with the jena API now (which looks like it's using SPARQL Update), so maybe it will be able to handle them correctly
<awoods>escowles: I was thinking that you were just going to listen for events from the repository, and update the external triplestore accordingly.12:31
...which could then be SPARQL queried.
<escowles>awoods: yes, that's what i'm doing -- except that updates require deleting the old data first12:32
<awoods>escowles: ok, and that is where you are hitting the issue: removing data from the external triplestore.12:33
<escowles>yes, i can remove the top-level data, but blank nodes get left behind
<awoods>escowles: When do we come across "blank nodes"?12:34
<escowles>awoods: any time we have child nodes with structure, such as fixity results12:35
<awoods>escowles: thanks for the status... and good luck with jena12:38
* ajs6f joins
blank nodes, by definition, can't be addressed by identifier.
only by constraint.12:39
Unless we can be sure that the SPARQL update in question uniquely identifies a particular node, we can't ensure that we will act on the nodes in question.
<escowles>ajs6f: yes, and i can imagine parsing the existing data, and building a bunch of SPARQL delete queries… but that would be error prone
<ajs6f>But for a wholesale delete, shouldn't child nodes get deleted automatically?
(by virtue of hierarchy)12:40
<escowles>ajs6f: if you have reasoning enabled, then yes they should
or if if you are using named graphs, you can just delete the whole graph
<ajs6f>No, I mean in the JCR.
Oh, wait— you're trying to update the _triplestore_.
I get it.
<escowles>oh yes, updating the triplestore based on JMS events
<ajs6f>This is why we thought so much about using named graphs per resource.
This is _exactly_ why.12:41
<escowles>yep, it would make updates very easy and queries fiendishly difficult
<ajs6f>Not if we overlaid it with a union graph.
(Which most triplestores can do, tho' not all.)
I'm not arguing for it.
Just pointing out that we've been here before.12:42
Maybe we can use UUIDs? Each node has a UUID. If that were attached to the RDF, maybe we could use it to do the deletes...12:43
<escowles>i thought about adding dummy subjects of some kind to all the blank nodes before updating the triplestore, the UUID could be a good thing to use for this
<ajs6f>yeah, it's managed for us, and it fulfills the conditions we'd want.12:44
<escowles>so they would be http://host:8080/rest/objects/abcd#15FEB9EB-3C04-4640-A074-ACE3323D3202
<ajs6f>http://host:8080/rest/objects/abcd#15FEB9EB-3C04-4640-A074-ACE3323D3202 hasUniqueId http://host:8080/rest/objects/abcd#15FEB9EB-3C04-4640-A074-ACE3323D3202
http://host:8080/rest/objects/abcd#15FEB9EB-3C04-4640-A074-ACE3323D3202 hasUniqueId http://host:8080/rest/objects/abcd#15FEB9EB-3C04-4640-A074-ACE3323D320212:45
Bad cut and paste.
But do you see what I mean? Resource hasUniqueId Resource#UUID
Or even as a literal:
Resource hasUniqueId "UUID"12:46
<escowles>yes: _:bn1234 hasUniqueId http://host:8080/rest/objects/abcd#15FEB9EB-3C04-4640-A074-ACE3323D3202
I don't know whether we might gain some efficiency with it as a literal.
Probably depends on the triplestore.
<escowles>OK, I'll see if Jena can handle deleting the blank nodes without that, but using the UUID seems like a possible way to handle them if not12:47
<ajs6f>Anyway, it would allow to construct SPARQL updates that kill discriminately.
Good luck.
* ajs6f leaves
* jay leaves12:49
<awoods>escowles: If you want the object (UUID) to be dereferencable, I believe the notation is from the root, i.e. http://host/rest/\[uuid\]
<escowles>awoods: good to know12:51
* jay joins13:02
<nbanks>awoods: Thanks!13:18
* github-ff joins13:20
[tuque] nigelgbanks pushed 16 new commits to fedora4: http://git.io/8d2Fsg
tuque/fedora4 ccace41 Jonathan Green: Break out Fedora 3 functions...
tuque/fedora4 674563e Jonathan Green: Changed symantics of HttpConnection to allow more parameters.
tuque/fedora4 e6a6096 Jonathan Green: Removed some crufty fcrepo4 stuff.
* github-ff leaves
* jongibson leaves13:30
* jongibson joins14:24
* jongibson leaves15:31
<escowles>awoods: i think i do need to get the UUID for the FixityResults so we can delete them -- but i don't see how to get that from FixityResult or LowLevelCacheEntry. any ideas?15:41
<awoods>escowles: I do not have a good idea of the top of my head. After my next meeting, I will give it a look.15:44
<escowles>awoods: thanks -- i'll post to the ff-internal list in case anybody else can get to it before then15:46
* jongibson joins15:56
* nbanks leaves16:14
* nbanks joins16:47
* nbanks leaves16:52
* nbanks joins17:47
* jay leaves17:50
* github-ff joins
[tuque] yqjiang opened pull request #4: Fedora4 (1.x...fedora4) http://git.io/02C5ug
* github-ff leaves
* ksclarke leaves17:51
* nbanks leaves17:53
* jongibson leaves18:45
* nbanks joins18:48
* ksclarke joins18:52
* nbanks leaves18:53
* nbanks joins19:18
* nbanks leaves19:23
* ksclarke leaves19:56
* nbanks joins20:18
* nbanks leaves20:23
* nbanks joins21:19
* nbanks leaves21:24
* nbanks joins22:19
* nbanks leaves22:24
* ksclarke joins22:33
* nbanks joins23:19
* nbanks leaves23:24

Generated by Sualtam