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

Using timezone: Eastern Standard Time
* eddies leaves01:26
* eddies joins01:27
* eddies leaves
* eddies joins
* kaarefc joins03:39
* github-ff joins03:57
[fcrepo4] eddies pushed 1 new commit to gh-pages: http://git.io/J3T6nw
fcrepo4/gh-pages 6408b72 Edwin Shin: Adds 3.6.3-SNAPSHOT site documentation
* github-ff leaves
* github-ff joins03:58
[fcrepo4] eddies pushed 1 new commit to gh-pages: http://git.io/cTi2qA
fcrepo4/gh-pages aa6de24 Edwin Shin: Adds 3.6.3-SNAPSHOT link to main page
* github-ff leaves
* github-ff joins04:08
[fcrepo4] eddies pushed 1 new commit to gh-pages: http://git.io/tFKoDw
fcrepo4/gh-pages 26ab52f Edwin Shin: Adds docs.fcrepo.org CNAME
* github-ff leaves
* eddies leaves04:24
* eddies joins04:42
* eddies leaves
* eddies joins
* github-ff joins04:53
[fcrepo4] eddies pushed 1 new commit to master: http://git.io/NNjzpw
fcrepo4/master d6ad261 Edwin Shin: Merge pull request #24 from futures/multipart-mixed-responses...
* github-ff leaves
* github-ff joins04:54
[fcrepo4] eddies deleted multipart-mixed-responses at 819a1a0: http://git.io/ZHuXVw
* github-ff leaves
* travis-ci joins05:04
[travis-ci] futures/fcrepo4#138 (master - d6ad261 : Edwin Shin): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/3e1330809885...d6ad26184a0a
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5066336
* travis-ci leaves
<ff-pivotal>Edwin Shin added "Switch Drupal multisite for Islandora to use fcrepo.org" https://www.pivotaltracker.com/story/show/4513821705:45
Edwin Shin added "Setup Jenkins to automatically deploy Maven snapshots" https://www.pivotaltracker.com/story/show/45138241
* eddies leaves05:57
* ppound joins07:37
* JasonDGI joins08:21
* VincentNG joins08:34
<cbeer>VincentNG: jenkins is at http://ci.projectblacklight.org/jenkins/view/FF/09:54
but we're using travis-ci for our CI builds
jenkins is just for deploy (or, would be, if it worked), and apparently to deploy mvn snapshots
<VincentNG>:-)09:55
thanks
<cbeer>Maven 2 Project Plugin: "Now supports Maven 3 too"09:57
:/
"To the best of my knowledge, this is almost never a useful option especially for CI servers like Jenkins."09:58
enabled by default
* barmintor joins10:03
<cbeer>yup, i remember why i hate jenkins again10:04
"[ERROR] uniqueVersion == false is not anymore supported in maven 3"10:21
even better.
<ff-pivotal>Chris Beer started "Setup Jenkins to automatically deploy Maven snapshots" https://www.pivotaltracker.com/story/show/4513824110:26
Chris Beer finished "Setup Jenkins to automatically deploy Maven snapshots" https://www.pivotaltracker.com/story/show/45138241
Chris Beer started "Multipart POST vs POST body content stream negotiation for add/modify datastream" https://www.pivotaltracker.com/story/show/45062931
Vincent Nguyen added comment: "related to https://www.pivotaltracker.com/story/show/45138241" https://www.pivotaltracker.com/story/show/4473343310:30
<cbeer>VincentNG: well, you're in luck
<VincentNG>:-)10:31
cbeer: need any help? extra set of eyes?10:34
<cbeer>just a sec. banging out code so I have something to report at standup10:35
* github-ff joins10:36
[fcrepo4] cbeer created multipart-form-data-datastream-upload (+1 new commit): http://git.io/Ce6zLw
fcrepo4/multipart-form-data-datastream-upload 68c00c5 Chris Beer: add multipart/form-data support for add/modify datastream API endpoints
* github-ff leaves
<ff-pivotal>Chris Beer added comment: "https://github.com/futures/fcrepo4/pull/27" https://www.pivotaltracker.com/story/show/45062931
Chris Beer edited "Multipart POST vs POST body content stream negotiation for add/modify datastream" https://www.pivotaltracker.com/story/show/45062931
<cbeer>ppound: i made you the requestor of https://www.pivotaltracker.com/story/show/45062931
i think it should make tuque happy10:37
<ppound>k we have a put function in tuque as well that i'm now using instead the post10:38
we have ingest and add data stream working properly now on top of fcrepo4
<cbeer>ppound++10:39
VincentNG: here are the two big things on my plate right now:10:43
- getting this stupid clustering thing working (find out why it's not working in fcrepo4, but does in a stand-alone project, etc, etc)
- https://www.pivotaltracker.com/story/show/4480719310:44
sorry, wrong ticket
https://www.pivotaltracker.com/story/show/44807121
i think eddies' suggested JSR-223 may not be the right term10:45
but the idea is to make writing indexing rules easier
by letting people do it in not-java.10:46
we have https://github.com/futures/ff-jms-solr-indexing already
<ff-pivotal>Paul Pound finished "Expose repo stats in Islandora" https://www.pivotaltracker.com/story/show/44813349
<cbeer>(and a similar concept in https://github.com/futures/ff-indexer-fourstore)
and two projects that provide dependencies (for reasons that are not at all clear to me):10:47
https://github.com/futures/jms-client
https://github.com/futures/service-runner
and for the solr case, i think the existing ff-jms-solr-indexing code is probably... not production-ready
it probably should use http://wiki.apache.org/solr/Solrj10:48
and do sensible things with spring
etc
so, if by "need any help" you're interested in that issue .... :)
<ff-pivotal>Chris Beer added comment: "i think eddies' suggested JSR-223 may not be the right term10:50
7:45
but the idea is to make writing indexing rul..." https://www.pivotaltracker.com/story/show/44807121
Paul Pound added comment: "Ingest, add datastream, purge datastream, purge object, viewing datastream content are working against fcrpo4..." https://www.pivotaltracker.com/story/show/4449071910:53
<cbeer>ppound: we don't have any notion of object labels in fcrepo4 yet10:54
* jcoyne joins10:58
* eddies joins
* eddies leaves
* eddies joins
<cbeer>(i just joined)11:01
eddies: i suggested those indexer tickets.11:03
+1
<ff-pivotal>Edwin Shin edited "fcrepo4 kitchen sink project" https://www.pivotaltracker.com/story/show/4473343311:04
* github-ff joins11:07
[fcrepo4] ajs6f pushed 1 new commit to master: http://git.io/t0Vbow
fcrepo4/master 13b5a8f ajs6f: Added Javadocs
* github-ff leaves
<ff-pivotal>Chris Beer added "Send cbeer CI/Sonatype artifact deploy credentials." https://www.pivotaltracker.com/story/show/45157493
Chris Beer edited "Send cbeer CI/Sonatype artifact deploy credentials." https://www.pivotaltracker.com/story/show/45157493
<cbeer>ppound: gsearch is tied up in SOAP11:08
that's an even bigger problem.
* github-ff joins11:10
[fcrepo4] ajs6f pushed 1 new commit to master: http://git.io/E9EM1w
fcrepo4/master 64d262f ajs6f: Added Javadocs
* github-ff leaves
<cbeer>hey, a standup that happened in a reasonable amount of time11:12
<VincentNG>is there a list of the merritt apis?
<cbeer>VincentNG: here's what I know:11:13
there's this storage api: https://wiki.ucop.edu/display/Curation/Storage
and we have an almost-complete implementation of that in the fcrepo-merritt project
but it has an awful add/update API involving multipart/form-data, a manifest, and some other nasty things11:15
there's this:
https://wiki.ucop.edu/display/Curation/Ingest
which I haven't looked at in depth, but i believe is an enhancement on top of the Storage API
<VincentNG>ok, let me take a look11:16
<cbeer>and then their actual implementation https://bitbucket.org/merritt/mrt-storecloud/
which has some of the storage API in it, and maybe the ingest api, and maybe some other stuff
(and what the other stuff is, i'm not sure yet)
(the specs also say a lot about how files should exist on the filesystem, which, for now at least, we're ignoring)11:17
hm. where did the merritt code go11:18
<VincentNG>this? https://bitbucket.org/merritt/mrt-core
<cbeer>https://github.com/futures/fcrepo4/pull/311:19
sorry, our merritt code
<VincentNG>oh
<cbeer>i guess it's just hanging out as a PR
* travis-ci joins
[travis-ci] futures/fcrepo4#139 (master - 13b5a8f : ajs6f): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/d6ad26184a0a...13b5a8fad61b
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5074441
* travis-ci leaves
<cbeer>https://github.com/futures/fcrepo4/blob/1146fa78342aa6a513413358f7377e0a999d8e83/fcrepo-merritt/merritt-fedora-mapping.txt
there's also that document
which maps merritt APIs to the fcrepo 3.x API11:20
<VincentNG>ahh. thanks
* travis-ci joins11:22
[travis-ci] futures/fcrepo4#140 (master - 64d262f : ajs6f): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/13b5a8fad61b...64d262f1cc6e
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5074490
* travis-ci leaves
* github-ff joins11:29
[fcrepo4] ajs6f pushed 1 new commit to master: http://git.io/pMxNtA
fcrepo4/master ed29128 ajs6f: Added even more Javadocs
* github-ff leaves
<JasonDGI>if i want to add another node.property to FedoraObject, are there any restrictions on how it can/cant be named?11:33
<cbeer>JasonDGI: you need to register it in the CND
<JasonDGI>CND?
<cbeer>https://github.com/futures/fcrepo4/blob/master/fcrepo-webapp/src/main/resources/config/fedora-node-types.cnd11:34
that's what defines what fields nodes that mixin types are allowed to have
* ajs6f joins11:36
<cbeer>ajs6f: want to reject this PR? https://github.com/futures/fcrepo4/pull/2711:37
<ajs6f>Eh? Is this just for legacy compatibility?11:38
<cbeer>ajs6f: yeah. and probably a bad idea11:40
<ajs6f>You could add it as a property on fedora:object, or on fedora:resource, if it's something that a datastream could also have (like label, for example).
Well, other than Islandora, are we expecting other people to expect it? (Hoenstly, I didn't even know fcrepo3 could do that.)11:41
<cbeer>i don't know. it was a surprise to me too
the use case, as best i can figure, is someone wanted to hook up an HTML form that went straight to fcrepo 3?11:42
<ajs6f>Urg.
it ain't that kind of framework, really.
I say leave it out for now (if the Islandora guys are cool with that) and let someone file a feature request (if anyone ever does).11:43
But then I'm a hard, hard man.
* travis-ci joins
[travis-ci] futures/fcrepo4#141 (master - ed29128 : ajs6f): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/64d262f1cc6e...ed2912800903
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5075247
* travis-ci leaves
<ppound>ajs6f: we should be ok if it's left out11:44
<ajs6f>Groovy. (The state of being, not hte anguage.)
<JasonDGI>are we talking about multipart forms or object labels?
<ajs6f>There are two conversations.11:45
Object nbdes, I'm pretty sure, can support labels now.
<JasonDGI>ajs6f cbeer you are two very confusing people
<ajs6f>We just need a method or two on FedoraObject/
You can look at the methods getLabel() and setLabel() on Datastream to see how to do it.
<JasonDGI>what about on FedoraObject, is that supported?11:46
i dont see it
<ajs6f>Er- that's what I just said.
Object nodes are ready to store it.
You'll have to add the code to FedoraObject to use that capability.11:47
And Datastream has examples you can see.
lines 89 and 106.
For lack of any better predicate, we're storing labels in dc:title, for now.11:48
* ppound leaves
<JasonDGI>so you never assert any kind of uniqueness for your labels?11:49
<ajs6f>No one asked me to.
You can change the CND schema to enfore that.
give dc:title on fedora:resource any constraint you like.11:50
<cbeer>(uniqueness for labels?)
<ajs6f>If we have unique labels, we shouldn't use dc:title. that's not the semantics of dc:title/11:51
<cbeer>i guess i should ask what we mean by unique labels.. just n=1 to match the fcrepo3 meaning?11:54
<ajs6f>That's how I took it.
Perhaps with a default value of the nodename or something like that.
afkvbrb11:58
* ppound joins
* ajs6f leaves12:02
* ajs6f joins12:25
* VincentNG leaves12:32
<eddies>ok, i'm off my calls…are folks still on for a call in ~25mins to talk APIs?12:34
<ajs6f>Only if we use a flaky conf line that knocks half of us off and doesn't let the other half on.12:35
<eddies>i'm happy to do a google hangout
<ajs6f>too reliable.
<eddies>we can take turns hanging up on each other to simulate
<ajs6f>:)'12:36
<eddies>adam, can you join in, too?
ack. what happened to vincent?
hrmph
<ajs6f>Sure, but I might be a little late
Are talking legacy or El Dorado or both or something else entirely?
<eddies>el dorado? qu'est que c'est?12:37
<ajs6f>The legendary API of gold.
Often sought, never found.
<eddies>oops. bad spelling. anyway, yes, el dorado ;-)
<ajs6f>Coders have died trying to get there. Will we be the next?12:38
Or will we return laden with rich abstractions and gleaming workflows...?
<eddies>well, more concretely, next steps for implementing some of the merritt apis and moving off the legacy-apis
<cbeer>(or, rabble-rousers will suggest the merritt apis just aren't worth it.)12:39
<ajs6f>Moving our effforts, or moving the community, or both..
?
I guess we have to do both if we do either.
<cbeer>(maybe the rabble-rousers should forget about joining the call)
<ajs6f>I think my rabble is already roused.
<eddies>rabble-rousers need to propose concrete alternatives
otherwise, it's just whinging
<cbeer>eddies++12:40
<eddies>heh. that or it's going to end up being CMIS ;-P
<ajs6f>As long as it's entirely based on SOAP and XACML.12:41
<eddies>yeah, i think CMIS will be our stick
merritt may not quite be a carrot, though
<ajs6f>I can see Eddie sitting on a dev's chest, screaming "You want CMIS!? Is THAT what you want!?"
<jcoyne>FC4 will support Corba, right?12:42
<ajs6f>No, we went with Cobra. It's a little-known alternative with venomous protocols.12:43
<eddies>http://cdn.memegenerator.net/instances/400x/35427837.jpg12:44
* VincentNG joins
<ajs6f>"Fedora: we cram it down your throat."
<JasonDGI>i keep on turning to this window when ajs6f says something crazy like that
<ajs6f>"Fedora: it's okay to be scared."12:45
<eddies>i think i like this better: http://cdn.memegenerator.net/instances/400x/35427906.jpg
<JasonDGI>"Fedora: please don't hurt me"
<ajs6f>"Fedora CMIS: Our requirements include face-stabbing."12:46
<eddies>i just like the idea of the field of battered and bloodied bodies upon which our apis have been forged ;-)12:47
<ajs6f>As long as they aren't ours.
<eddies>of course not, we are those who remain12:48
<ajs6f>"Fedora: you will know it by the trail of dead."
<eddies>ajs6f: i couldn't get enunciate to work with the maven site generation, btw12:49
<ajs6f>cbeer? (He brought enunciate into the game. I just thought it sounded good, and if my friends jump off a bridge, so do I.)
<eddies>(for docs-only config, not for bringing up the endpoint)12:50
oh. i thought that was you
<cbeer>i know nothing.
<ajs6f>I thought we only had it running for docs.
<jcoyne>ajs6f http://xkcd.com/1170/
<cbeer>i just wanted rest api docs
<ajs6f>Amen to that.
Although (getting back on the old tired soap-box) I still claim we should do a set of HTML responses for the REST API that provide simple management.12:51
<eddies>well, in any case. it wasn't working for me when i tried to configure it as part of the site generation
ajs6f++ i'm still in favor of that
<ajs6f>That also might be a place for folks who can't offer heavy dev rsources to contribute.
<eddies>i'd actually like to see a working PoC of that12:52
that uses content-negotiation
<ajs6f>My favorite example is Apache Stanbol. I can do a simple example on /rest/objects/, if you want.
<eddies>i'd like any new stuff not to happen on the legacy-api module, tho12:53
<ajs6f>Okay, I'll do it on the _other_ API. :)
(CMIS)
<eddies>well, that's why we're talking in 6.5mins =)
<ppound>I'm running a bit late
<eddies>we'll just start assigning you tickets that are due tomorrow for every minute you're late ppound =)12:54
cbeer: where is the merritt work you started on in chicago?12:55
<ajs6f>afk brb
* ajs6f leaves
<cbeer>eddies: in a pull request
<eddies>https://github.com/futures/fcrepo4/pull/3
?
<cbeer>sounds right12:56
<eddies>travis seems unhappy with that branch: https://travis-ci.org/futures/fcrepo4/builds/475334912:58
<cbeer>yep.12:59
that's the whole multipart mess i got into
i think i figured it out in master now though
* ajs6f joins
<JasonDGI>are we on freeconf?13:00
<eddies>i was trying to set up a hangout
do i have to manually invite folks?13:01
i was hoping for a url to share
<ajs6f>You should be able to go the Futures "community and trigger a hangout there.
<eddies>i did
<cbeer>https://gist.github.com/cbeer/5c9fa3a6c9246446723d
there's the APIs the mrt-storecloud hs13:02
has
beats me where some of them came from.
<eddies>can folks visit https://plus.google.com/u/0/communities/110386469911510107748
and see if they can join the hangout?
<ajs6f>I went to the Futures comminnt and there's a button to join the hangout.
* barmintor is in the hangout13:03
<eddies>vincentng are you joining us?13:04
<barmintor>so far, this is better sound than I get from FREECONFERENCECALLHD
<JasonDGI>i sent a join request13:05
<eddies>i approved
<VincentNG>ya
<JasonDGI>how do i get in the call?
<VincentNG>i don't see a hangout request
is anyone on the call?
<ajs6f>Only cool people.
Go to the link Eddie sent.13:06
And click on Eddie's hangout.
<VincentNG>i don't see "Eddie's hangout", but i think my join request needs to be approved
<cbeer>VincentNG: approved.13:07
http://support.google.com/chat/bin/answer.py?hl=en&answer=252061413:08
<VincentNG>i'm in
but i don't think i have a mic13:09
*signing*
<JasonDGI>how were you talking during the previous meeting?
<VincentNG>JasonDGI: phone13:10
<cbeer>ah, i found the telephone button
VincentNG: "Invite People" -> "+telephone" -> "enter number"13:11
<VincentNG>cbeer: thanks13:15
<cbeer>https://wiki.ucop.edu/download/attachments/16744573/Merritt-ingest-service-latest.pdf?version=23&modificationDate=135732788300013:21
* ppound leaves13:24
<ajs6f>CMIS is on the table!13:25
Let's all beat on it.
<barmintor>What's SD's REST API?13:29
<ajs6f>SD?
<cbeer>UCSDs?
<barmintor>fedora_specific_api--
cbeer: yes
<cbeer>let me find the page13:30
https://github.com/ucsdlib/dams/wiki/REST-API
<JasonDGI>paul's here13:33
* ppound joins13:40
<barmintor>"Simpler than SOAP!"13:43
"Faster than Glacier!"
the lack of consistency in responses, lack of content negotiation, lack of repository actions, and APIM/APIA divide are all frustrating13:44
<cbeer>PATCHES WELCOME!13:45
https://github.com/hedtek/dspace-rest !13:46
done.
<JasonDGI>i have a quick meeting with mark in 10 minutes, ill have to duck out soon13:53
<ajs6f>http://www.xaop.com/labs/activecmis
<cbeer>yeah, i was just looking at that13:54
the lack of tests is.. not so inspiring13:55
<ajs6f>PATCHES WELCOME!
:0
<cbeer>ok, i need to drop. got to get to the office for lunch.13:57
<barmintor>http://roy.gbiv.com/untangled/2008/no-rest-in-cmis13:59
<ajs6f>I was just reading that.
Roy must be fun at cocktail parties.
"This cocktail weenie is not RESTful!"14:00
"This martini is not hypertext driven!"
<jcoyne>this martini should tell me where I can get the next one.14:03
<ajs6f>After the third or fourth, it will.14:04
<barmintor>the DAMS REST api is a prety fair iteration on the FCR3 API.14:09
<ajs6f>Did they not lose all the relationship management stuff?14:10
(Intentionally...)
<barmintor>the FCR3 rel API sucks
<jcoyne>ajs6f: their metadata is all RDF
<barmintor>https://github.com/ucsdlib/dams/wiki/REST-API
<ajs6f>I agree, but I think we need _some_ way to manipulate relatinships.
<jcoyne>ajs6f: so the relationships are indexed by Hydra into solr
<barmintor>I agree with that
<ajs6f>jcoyne: sure, but it's exposed _as_ RDF.14:11
That's not terible or wrong, but it is different from what Fedora has traditionally done.
<jcoyne>ajs6f: Hydra doesn't really care about where the relationships are stored.
<ajs6f>That's fine. For Hydra.
<jcoyne>right;
<barmintor>ajs6f: http://www.w3.org/TR/2013/PR-sparql11-http-rdf-update-20130129/
<jcoyne>and Hydra-dams is hydra, thus it doesn't care about relationships
<ajs6f>But UVa does. APTrust does. Lots of folks do.14:12
Without relationships between objects, Fedora is an incredibly expensive file system.
With the minimal additional semantics ofthe obejct model.
<jcoyne>ajs6f: not arguing with you.
<ajs6f>I'm all in favor cleaning up the managment of relationships.14:13
https://wiki.duraspace.org/display/FF/ModeShape+CMIS14:16
https://github.com/ModeShape/modeshape/tree/master/web/modeshape-web-cmis-war
https://github.com/ModeShape/modeshape/tree/master/web/modeshape-web-cmis14:17
<jcoyne>ajs6f: how do you use relationships in fedora at UVA? Do you have a simple use case?
<ajs6f>Simple or complex.
We use them to model content all over the place.
<jcoyne>ajs6f: but do you need to use an api to do that? Can't you just write that into RELS-EXT?14:18
<ajs6f>E.g.
We could do it any number of ways. We cold do it in a SQL db alongside Fedora. :)
And for aI'm not against managing rlationships via manifest.14:19
manifests.
<jcoyne>ajs6f: But I'd like to understand why it's better to make fedora manage relationships, instead of just storing your relationships in fedora.
<ajs6f>It's not inherently better, it's easier for us.14:20
Changing one relationship is simpler and easier than pulling a massive chunk of RDF?XML, parsing it, making a change, then reprersisting it.
Even tho' the Fedora APIs are indisputably ugly.
<eddies>https://repository.jboss.org/nexus/content/groups/public/org/modeshape/modeshape-web-cmis-war/3.1.2.Final/modeshape-web-cmis-war-3.1.2.Final.pom14:21
<ajs6f>I think this may look differently depending on how many relatinships an object participates in.
<jcoyne>ajs6f: I would think the api would be ungainly for more complex cases (like a qualified relationship)14:22
<ajs6f>Perhaps that turns out to be a matter of taste. :)
<jcoyne>So you say: my_obj hasCreator to a B-node, which has a creator assertion pointing to a pid and a role assertion which is "Author"14:24
?
<ajs6f>And twenty or thirty other assertions at least.
On every object.
Not to mention RELS-INT.
We model very atommically.14:25
E.g.
<jcoyne>That's fine, I just was unaware there was an API to support that.
<ajs6f>Images and texts make up pages make up a book.
The relationship API? Yeah, its there. And as Ben says, its not very good.
And I'm all in favor of a better alternative.
http://www.w3.org/TR/2013/PR-sparql11-http-rdf-update-20130129/ could very well be that.14:26
<jcoyne>ajs6f: that begs the question, if I have subjective metadata, and I can model it in RDF, why not put everything in there?
<ajs6f>You should.
Although I'm not sure what you mean by "subjective"
.
<jcoyne>e.g. book title
not relationship data
<ajs6f>So description?
<jcoyne>sure
<ff-pivotal>Edwin Shin added "Update kitchen-sink-webapp to include CMIS support" https://www.pivotaltracker.com/story/show/45175517
<ajs6f>Sure, absolutely put it in RDF>
<jcoyne>ajs6f: by using the relationships api?14:27
<ajs6f>Then you can use different vocabularies together instad of being limited to one XML document <-> one format
Or by maniupulating the RDF directly, if that feels better.
We use the API, but we could use RELS-*.14:28
<jcoyne>ajs6f: interesting, I didn't know that was possible with Fedora.
<ajs6f>It's one of Fedora's oldest and strongest qualities— a real commitment to the graph.
<barmintor>ajs6f: ehh… sort of.14:29
<ajs6f>Well, at least as old as the 2-series
<barmintor>I think the only graph commitment FCR3 *realy* makes is the "an object is like a named graph of datastream descriptions"14:30
<ajs6f>I'm not talking about the codebase so much as the praxis around it.14:31
Finely-grained modeling is, to my knowledge, the norm.
<barmintor>right, but then it's not FCR3's commitment to the graph- it's yours!
<ajs6f>It's the Fedora community's comittmet of practice.
<barmintor>ajs6f:I wonder how much that's the case.
<ajs6f>And the community is far more important than the code bae.14:32
<barmintor>(fine grained modeling)
<ajs6f>Well, I don't have exactl fgures. :)
<barmintor>:)
<ajs6f>Can you think of someone doing compound modeling?
<barmintor>ajs6f: we just zoomed past each other14:33
I meant I think the cmodels are fairly generic
though we're actually doing more compound modeling, in a sense
<ajs6f>I'm not talking about cmodels as much as about the granularity of modeling.14:34
<barmintor>since it was infuriating fetching derivative bitstreams by going back to the RI
<ajs6f>the lack of consistency in responses, lack of content negotiation, lack of repository actions, and APIM/APIA divide are all frustrating
* github-ff joins14:47
[fcrepo4] ajs6f pushed 1 new commit to master: http://git.io/dnkLWA
fcrepo4/master 299afef ajs6f: Added fcrepo-htt-api module
* github-ff leaves
<cbeer>clustering is going to drive me MAD.14:50
i think it's.. working
<ajs6f> IT'S ALIIIVVVE!!!14:51
<cbeer>i don't think i'm doing ANYTHING different than before
must be thinking very positive thoughts or something
* ppound leaves
* ppound joins
<cbeer>oh, oops. i turned off eviction14:52
i think that was a known issue (supposedly fixed in ISPN, but MODE hasn't confirmed)
and i've long suspected it's actually still broken
<ajs6f>What issue?
<cbeer> https://issues.jboss.org/browse/MODE-1754
that's the MODE ticket
but it's linked to the (fixed) ISPN ticket14:53
hm.14:55
or i'm also seeing https://issues.jboss.org/browse/MODE-183014:57
<ff-pivotal>A. Soroka added "Provide examples of HTTP conneg for fcrepo-http-api" https://www.pivotaltracker.com/story/show/4517870515:04
A. Soroka edited "Provide examples of HTTP conneg for fcrepo-http-api" https://www.pivotaltracker.com/story/show/4517870515:05
A. Soroka started "Provide examples of HTTP conneg for fcrepo-http-api" https://www.pivotaltracker.com/story/show/45178705
* travis-ci joins15:10
[travis-ci] futures/fcrepo4#142 (master - 299afef : ajs6f): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/ed2912800903...299afefa7380
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5081341
* travis-ci leaves
* VincentNG leaves15:11
* ajs6f leaves15:29
* VincentNG joins
* ajs6f joins15:31
Any suggestions for where to hang the "http api" on the URL tree? 'rest' is already taken by the legacy api module.15:32
<jcoyne>ajs6f: /15:34
<ajs6f>We had some reason for not doing that with the legacy API, but I can't remember what it was. cbeer, do you remember?15:35
<jcoyne>ajs6f: or to be future proof how about /v4/
<barmintor>how about 'rest', and if you have to have the legacy api there you can configure something else15:36
<ajs6f>Okay. I didn't want to do that, even tho' it's the right thing, because now I'm going to have to rewrite parts of the tests. Well, who wrote them that way to begin with?! Oh, yeah. Me.15:37
<barmintor>sorry :P
and thank you!
<ajs6f>And the worst part is, I'm not even going to learn anything from it.15:38
<jcoyne> /v3/ for the old rest api
<barmintor>jcoyne++, tho you have to be able to wire it in at the 'rest' endpoint
OR DO YOU?!
<jcoyne>mod_proxy++
<ajs6f>Wait, so have we come to /rest/v3 and /rest/v4?
<barmintor> /v3 and /rest15:39
<jcoyne>sounds good
<ajs6f>Is that what we want? +1 to vote for it.
Oooh….
Okay, /v3 and /rest— gimme a −1 or I go with it.
<barmintor> /your_problem and /rest
<ajs6f>How about /update-your-crap and /moving-target?15:40
<barmintor> ajs6f++ # winner
<ajs6f>Okay, /v3 and /rest it is.
Wait, I'm an idiot. I don't have to update the tests, because they only load one API at a time.15:41
Cool.
<jcoyne>ajs6f++15:42
<ajs6f>Saved by my own laziness!
* github-ff joins15:51
[fcrepo4] ajs6f pushed 1 new commit to master: http://git.io/nrH9zA
fcrepo4/master 80dcb7e ajs6f: Added new HTTP API module to fcrepo-webapp
* github-ff leaves
<ajs6f>Looks like CXF only offers two ways of providing HTML views: XSLT and JSP. (Nuts— RESTEasy lets you use a bunch of different nicer templating languages!) Any preference in the team?16:03
Keeping in mind that this is just to get some examples out there.
Okay, I'm going with XSLT. We can use something else later if we want to.16:06
* ppound leaves16:08
* jcoyne leaves
* travis-ci joins16:09
[travis-ci] futures/fcrepo4#143 (master - 80dcb7e : ajs6f): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/299afefa7380...80dcb7e464ca
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5083203
* travis-ci leaves
* JasonDGI leaves
* jcoyne joins16:11
<cbeer>ajs6f: what's the plan with the new API? make it up as we go along?16:12
prove to eddies we can design a pleasing rest api?
<ajs6f>Right now, just starting with legacy and adding HTML responses.
Just to prove to me that we can do that.16:13
Going forward, you tell me.
:)
* VincentNG leaves16:20
* ajs6f leaves
* ajs6f joins16:23
* kaarefc leaves16:49
* github-ff joins16:51
[fcrepo4] ajs6f pushed 1 new commit to master: http://git.io/pAdH8A
fcrepo4/master fa8191f ajs6f: Added first simple conneg example at /rest/objects/{pid}
* github-ff leaves
<ajs6f>o and o for the day. Further conneg examples tomorrow.
* ajs6f leaves16:56
* travis-ci joins17:03
[travis-ci] futures/fcrepo4#144 (master - fa8191f : ajs6f): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/80dcb7e464ca...fa8191f1b5e8
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/5084711
* travis-ci leaves
<ff-pivotal>Chris Beer started "Extend ff-jms-solr-indexing to support JSR-223 scripting-based indexing" https://www.pivotaltracker.com/story/show/4480712117:46
Chris Beer edited "Create fcrepo4 <-> solr indexer that uses JSR-223 (or otherwise) to allow scripting language-based indexing" https://www.pivotaltracker.com/story/show/4480712117:47
* jcoyne leaves17:59
<cbeer>YAY! MODE 3.1.3 might fix part of our clustering trouble19:13
* jcoyne joins20:53
<eddies>cbeer: i presume you're referring to https://issues.jboss.org/browse/MODE-183022:33
<cbeer>eddies: yep22:35
i can't build MODE from source, or i'd check now
(probably have to set something up in m2/settings.xml to use the jboss repositories)22:36
<eddies>well, it's due out today, so shouldn't be long
<cbeer>rhauch said he'd have it out in the morning
(I will say, for as much pain as MODE/ISPN seems to cause, at least they're very responsive)22:37
<eddies>yes. but just imagine if we had expended this much effort on getting lily multi-node ;-)22:39
<cbeer>then we'd still be struggling with hadoop deployment?
<eddies>heh
* eddies leaves23:17
* eddies joins23:20
* eddies leaves
* eddies joins

Generated by Sualtam