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

Using timezone: Eastern Standard Time
* kaarefc joins02:22
* jonathangee leaves03:26
* mikeAtUVa leaves03:28
* jonathangee joins03:29
* mikeAtUVa joins03:30
* nbanks joins07:23
<pivotal-bot>Nigel Banks finished "UI glossary documentation" https://www.pivotaltracker.com/story/show/6183027407:50
Esme Cowles added comment: "The 1TB upload/download finished successfully. I've updated the wiki (https://wiki.duraspace.org/display/FF..." https://www.pivotaltracker.com/story/show/6177343007:52
Esme Cowles finished "Determine filesize ingest limit via REST API" https://www.pivotaltracker.com/story/show/61773430
Nigel Banks started "Create a Mac OSX Dock icon for single click launcher" https://www.pivotaltracker.com/story/show/6182473407:53
Nigel Banks added comment: "Found an interesting tutorial to make a full OSX application via a Maven plugin. http://simplericity.com/200..." https://www.pivotaltracker.com/story/show/6182473408:02
Nigel Banks added comment: "Hmm the tutorial doesn't seem useful as we don't have a main class, entry point for the generated war from t..." https://www.pivotaltracker.com/story/show/6182473408:12
Nigel Banks added comment: "http://grepcode.com/file/repo1.maven.org/maven2/org.simplericity.jettyconsole/jetty-console-core/1.42/JettyC..." https://www.pivotaltracker.com/story/show/6182473408:18
Esme Cowles started "Determine filesize read limit via REST API" https://www.pivotaltracker.com/story/show/6177364408:31
Frank Asseg added comment: "@ericjames It should be bossible to tell projected nodes from "normal" nodes so I guess it would make sense ..." https://www.pivotaltracker.com/story/show/5600291608:35
* ajs6f joins08:44
<pivotal-bot>Andrew Woods added comment: "That is great, @escowles." https://www.pivotaltracker.com/story/show/6177343008:49
Andrew Woods accepted "Fix kitchen-sink build" https://www.pivotaltracker.com/story/show/61874624
Andrew Woods accepted "Default Storage Parameter Directory: tmp to cwd" https://www.pivotaltracker.com/story/show/6100433608:50
Andrew Woods accepted "UI glossary documentation" https://www.pivotaltracker.com/story/show/61830274
Andrew Woods accepted "Determine filesize ingest limit via REST API" https://www.pivotaltracker.com/story/show/6177343008:51
<escowles>awoods: i've got two testing tasks: setting up 2-node cluster using gregjansen's setup, and testing serving large files from filesystem federation -- any preference for which i should do first?
<awoods>escowles: maybe the filesystem federation first.08:52
<escowles>ok, i'll focus on that first
<nbanks>Does any one have a real good tutorial they could recommend for maven? I'm trying to get the output of one plugin to be the input of another. Is it possible? So far the only leads I have is to use another plugin like maven:assembly or maven:release?09:00
<ajs6f>nbanks: What exactly are you trying to do?09:01
<nbanks>ajs6f: I found this neat tutorial for making an OSX bundled application from a java app: http://simplericity.com/2007/09/19/1190234671861.html And I wanted to use it to address the pivotal ticket for making the OSX icon for the one click launcher. But I want it to use the jetty-console war not the fcrepo-webapp's jar.09:03
<ajs6f>nbanks: And you're trying to do this in fcrepo-webapp?09:04
<nbanks>ajs6f: Ya in the pom.xml
<ajs6f>nbanks: You're creating a subsidiary artifact. If the plugin you are using doesn't offer specific config for your need, then I know it will seem like extra paperwork, but you really want to have a new module for that. The new module will take a dependency of the jetty-console artifact and do whatever it does to wrap it as a Mac app. Or, you can use something other than the jetty-console plugin to provide a bundled Jetty or other server. I just picked that b09:06
<nbanks>ajs6f: OK, so where should I put this? Kitchen sink or another repository? Or another sub-project, in fcrepo4?09:10
* ermadmix joins09:11
<ajs6f>nbanks: I would put it as a separate project, e.g. fcrepo-mac-app or something. You could make it a submodule of fcrepo-webapp, but I think that would be confusing.
<nbanks>ajs6f: OK, I'll create a new repo, thanks!09:12
<ajs6f>nbanks: np
* kaarefc leaves09:19
<awoods>Welcome back, ajs6f. I hope you are feeling better.09:27
<ajs6f>ish
afk bbs
* ajs6f leaves
<nbanks>awoods: I'd bug adam, but he's not around. In maven is it possible to get an artifact that was generated from a parent pom.xml that isn't the primary artifact. I want to have my external project depend on the jetty-console.war but I am not sure how I can do so.09:35
<awoods>nbanks: yes
nbanks: generally, you can specify the <qualifier> in the <dependency> elements.09:36
<nbanks>awoods: oh thanks! I'll check that out
<awoods>nbanks: In this case, it looks like the <qualifier> is "jetty-console"09:37
nbanks: sorry, change "qualifier" to "classifier"09:38
<nbanks>awoods: I can't use qualifier but I can use classifier, (I found from searching online for qualifier)
awood: cool, thanks!09:39
* ksclarke joins09:46
* ajs6f joins09:57
* osmandin joins10:09
<pivotal-bot>Mike Durbin added comment: "There are tests that go against the API directly that cover (in a slightly different way) this functionality..." https://www.pivotaltracker.com/story/show/6147423610:17
Mike Durbin started "FedoraHtmlResponsesIT#testVersionCreationAndNavigation fails sometimes." https://www.pivotaltracker.com/story/show/6201622010:18
Andrew Woods edited "FedoraHtmlResponsesIT#testVersionCreationAndNavigation fails sometimes." https://www.pivotaltracker.com/story/show/6201622010:19
<barmintor>awoods: thought about it last night, and decided to change my vote re: event emit10:21
awoods: kernel service objects don't actually know when an event is "final"
<awoods>barmintor: what is your vote?
<barmintor>awoods: because of transaction rollback
awoods: so I need to come up with an idea about NodeRemoved10:22
<awoods>barmintor: Which means we must rely on the JCR events?
* barmintor starts thinking
<awoods>...which do not exist for NodeRemoved.
barmintor: Is the lack of Modeshape support for NodeRemoved events a JCR spec limitation, or a Modeshape under-implementation of JCR?10:23
<barmintor>awoods: It's not lack of support for NodeRemoved, it's lack of support for NodeRemoved-but-only-if-it-was-fedora:object/datastream10:24
<awoods>barmintor: Is it unreasonable to simply let all of the NodeRemoved events get pushed onto the eventbus, unfiltered?10:25
<barmintor>awoods: That's what I was referring to as "naked" NodeRemoved events yesterday, and what I am profiling this morning10:26
<ajs6f>They'll kind of have to get filtered at some point, or workflow like the indexer will get swamped with weird inappropriate messages. But they could be filtered downstream (outside of core).10:27
Like at the indexer.
<barmintor>ajs6f: I don't know how the downstream process would filter them10:28
the node is gone
it is an ex-node10:29
<ajs6f>Mm. Right.
<barmintor>it has ceased to be.
<awoods>ajs6f: That is the question I have, "Is it unreasonable to simply let all of the NodeRemoved events get pushed onto the eventbus, unfiltered"? meaning, will the indexer get swamped with "NodeRemoved" events.
<ajs6f>The ghost of the shadow of the absence of a node.
awoods; I know. barmintor is figuring out right now whether that is reasonable.
<awoods>ajs6f: I assume the indexer takes the NodeRemoved event and asks to remove the related entry from its index... if there is no entry to remove, so be it.10:30
<ajs6f>awoods: True, but rather inefficient. But it could work for now..
<awoods>ajs6f: on another topic...10:31
<escowles>could we tell what the ex-node was based on its path?
<barmintor>awoods: another JMS topic? HAHAHAHAHAHAHAHAHAH
escowles: alas, no
<ajs6f>Do people have the expectation that the index _is isomorphic to the repo, or that it is a homeomorphic transform of the repo?
<awoods>ajs6f: The question came up yesterday regarding the fixity transient URLs
<ajs6f>awoods: I saw. I don't care. We talked about it in IRC and I did what I was told to do. The nodes can be blahblah#foo or whatever we want.10:32
Just let me know what we want.
<awoods>ajs6f: I think the issue at present is that the URLs do not resolve.
<ajs6f>awoods: Then we'd better think about how the fixity triples are created. cbeer?10:33
<awoods>ajs6f: And since they are transient, I suspect they will never resolve.
<ajs6f>awoods: that's not illegal, just unpleasant.
Memento will solve all of these problems!10:34
<awoods>ajs6f: You mean unresolving URLs are not illegal?
<ajs6f>Right. They're just rude. Have you never gotten a 404 in your browser?
We can stick non-URL identifiers in there, instead.10:35
Whatever.
info:fedora:DONTLOOKATMEI'MNOTREALLYHERE/timestamp
<awoods>ajs6f: Maybe we can get fasseg and cbeer and others to stick on the line after standup to pick a direction.
<fasseg>ajs6f: awoods: that'd be great because I have a couple of questions about how to implement the fix for the federation fixity issue...10:36
* ermadmix leaves10:38
* ermadmix joins10:42
<barmintor>benchtool has abandoned me
* tecoripa joins10:45
* tecoripa leaves
* ajs6f leaves
<gregjansen>awoods: I found there was more work in progress code in the authz projects.. i.e. things that haven't made it into future with a PR yet. Test refactoring, code refactoring to enable reuse in the xacml pep, etc.. So I can either (1) create a big PR now since all ITs pass and changes to functional code are minimal (2) postpone this project migration into fcrepo of authz stuff (keep saving these commits separately) or (3) throw away this wo10:48
rk (probably not)
* nbanks leaves10:49
* mikeAtUVa leaves10:51
<pivotal-bot>Gregory Jansen added comment: "@awoods a PR based on this comparison would bring all code into master prior to migration of 2 sub-projec..." https://www.pivotaltracker.com/story/show/6201958610:52
* mikeAtUVa joins10:53
* ajs6f joins
<cbeer>ajs6f: re: fixity urls, i was surprised they weren't .../fcr:fixity#status/12345678 or something.10:54
resolvable and transient.
<ajs6f>we wanted them timestampes.
<cbeer>ajs6f: sure, but as part of the hash segment10:55
<ajs6f>Fine with me.
<cbeer>ship it.
<ajs6f>No, I'm fooling with this indexer stuff.
But we can change that at the standup.
* github-ff joins10:57
[fcrepo4] gregjan opened pull request #189: adding access roles endpoint to RDF lexicon (master...master) http://git.io/F5vrhQ
* github-ff leaves
<pivotal-bot>Gregory Jansen added comment: "https://github.com/futures/fcrepo4/pull/189" https://www.pivotaltracker.com/story/show/61850482
<awoods>gregjansen: Are there tickets for the other outstanding authz work?
<gregjansen>awoods: I think I entered them all, brb w/links10:58
* nbanks joins10:59
<gregjansen>awoods: well, they are several. just click on GJ if you want to see them all
<awoods>ok
<pivotal-bot>Gregory Jansen edited "Add access roles API to fcrepo-webapp" https://www.pivotaltracker.com/story/show/62019902
Gregory Jansen edited "move fcrepo-auth-roles-basic into fcrepo" https://www.pivotaltracker.com/story/show/62019734
<gregjansen>I will move them into the backlog, some were in icebox11:00
<awoods>All: standup
<gregjansen>the current question was about how to handle some commit outstanding in my old repo.. going to standup
<mikeAtUVa>https://plus.google.com/hangouts/_/calendar/eW91cm1lZGlhc2hlbGYuY29tXzVlYzdpNXQ2Z282dTdidHI4aTVrbGJxOTUwQGdyb3VwLmNhbGVuZGFyLmdvb2dsZS5jb20.opn3ai06k1ge0no877ei3cku4o
<awoods>cbeer? barmintor? standup11:02
<cbeer>awoods: there.
<barmintor>I continue to work on Events11:03
<escowles>synth_barmintor--
<ajs6f>Darkwave barmintor.
Glitchcore barmintor.
<barmintor>I need a good idea
a really good one
<ajs6f>Switch platorm. This ModeShape thing isn't working.
How about Lily?
<cbeer>you have 1 week.11:04
<ajs6f>We needed less than that last time.
<pivotal-bot>Chris Beer added comment: ""I removed the following line as session is already injected, and it worked past that line:" <-- is that real..." https://www.pivotaltracker.com/story/show/6194263811:08
<barmintor>I hvae been unable to get benchtool/master to run against fcr4, btw11:17
<pivotal-bot>Gregory Jansen finished "Bundle test workflow, platform, scale profiles into a cluster-hosted IT based on cargo" https://www.pivotaltracker.com/story/show/6182355411:24
Gregory Jansen added comment: "https://github.com/futures/fcrepo-test-profiles/blob/master/pom.xml" https://www.pivotaltracker.com/story/show/61823554
<ajs6f>cbeer: +111:31
<fasseg>the exception is: https://gist.github.com/fasseg/782777911:32
<ajs6f>Blank nodes are disallowed in serialization, but I'm not sure that has anything to do with this.11:33
<pivotal-bot>Osman Din added comment: "I updated the documentation for your review. As for the last bullet item, it seems to me that the serialized f..." https://www.pivotaltracker.com/story/show/4901279911:38
Osman Din finished "Introspect the bson output for modeshape schematics and report back how friendly the output is" https://www.pivotaltracker.com/story/show/4901279911:39
Frank Asseg added comment: "After talking on the hangout, we agreed on an abstraction level for CacheEntries, so we can pass a Projected..." https://www.pivotaltracker.com/story/show/5600291611:40
<awoods>Are others unable to use benchtool against F4 as barmintor notes?11:44
* github-ff joins11:46
[fcrepo-authz] gregjan opened pull request #4: Brings futures authz up to state of the art prior to git migration to fcrepo (master...master) http://git.io/xS3Y-w
* github-ff leaves
<barmintor>ajs6f: PropertyChanged/Added events- do we want to send a seperate event out for al the properties, or a single event indicating properties changed? which one should be in the legacy api?11:47
IIRC, legacy would be one event
<ajs6f>barmintor: Since we have no container for events, I suppose we should just stream them all. One per prop.
I think this is why JCR offers EventIterator and calls observers with it.
<barmintor>ajs6f: but we're measuring FCR4/Legacy against FCR3, so that's not helpful11:48
<ajs6f>Or we could compress them, as you say, but at what boundary?
<awoods>ajs6f: but we do not have specific details in the events which would give property-level info.
<barmintor>ajs6f: jcr:lastModified
<ajs6f>awoods: yes. So?
<awoods>ajs6f: so would all of the propertyUpdate events look the same?
in which, why not batch them.11:49
<ajs6f>barmintor: With a region? Like, within X time of each other, we consider these evnts to be bundle-able?
awoods: because we don't know if they actually are the effects of some single action the user would consider a single action.
<barmintor>ajs6f: If you didn't create or remove a node, and there was a Property event in the iterator, then we send jcr:lastModified
<ajs6f>That's why boundaries or containers are relvent11:50
barmintor: Meaning we let he user decide whether they belong together?
<barmintor>ajs6f: that forces FCR4 to adopt a performance impediment vis-a-vis FCR3
<ajs6f>barmintor: Eh? I don't know much about what FCR3 did.11:51
<fasseg>cbeer: ajs6f: we could even indicate the success ot the failure in the URI fragment of a fixity result? like .../fcr:fixity#data/success/HASH or /fcr:fixity#data/[0|1]{HASH} so you can immediately see what it was...11:52
or should the hash be created from distinct part itself?11:53
<cbeer>fasseg: i'd rather have opaque uris, myself.
<ajs6f>fasseg: That's seems like we're reducing the opacity of the identifier. How about a property on that identifier indicating the success or failure.
what cbeer said ^^^
<fasseg>not even in the fragmen?11:54
<ajs6f>Fragments are still part of the uri. {grin}
<fasseg>but I guess you're right it's part of the URI
* nbanks leaves11:55
<pivotal-bot>Frank Asseg added "Update produced fixity result URIs" https://www.pivotaltracker.com/story/show/6206329211:57
Andrew Woods added comment: "@gregoryjansen, The addition of the one HAS_ACCESS_ROLES_SERVICE constant does not address the inclusion of..." https://www.pivotaltracker.com/story/show/6185048211:58
Eric James added comment: "Yes it's true in that- 1) session gets injected, 2) sessions.getInternalSession() tries to redefine session,..." https://www.pivotaltracker.com/story/show/6194263812:00
Gregory Jansen added comment: "@awoods agreed, this ticket is not done." https://www.pivotaltracker.com/story/show/6185048212:01
<cbeer>infinispan-- # why do transactions need to be so hard?12:02
<pivotal-bot>Gregory Jansen added comment: "@awoods however, roles cannot currently be set at the root level. So I can add a roles form to the object..." https://www.pivotaltracker.com/story/show/61850482
Andrew Woods added comment: "That sounds like the right approach, @gregoryjansen." https://www.pivotaltracker.com/story/show/6185048212:03
<gregjansen>awoods: I do need to merge https://github.com/futures/fcrepo4/pull/189 before the authz PR will pass travis.12:07
<awoods>gregjansen: which ticket is that?12:08
<gregjansen>this RDF lexicon change in fcrepo is for https://www.pivotaltracker.com/s/projects/684825/stories/6185048212:09
<pivotal-bot>Gregory Jansen added comment: "https://github.com/futures/fcrepo-authz/pull/4" https://www.pivotaltracker.com/story/show/62019586
<awoods>gregjansen: done12:10
<gregjansen>awoods: thx!
* github-ff joins
[fcrepo4] awoods closed pull request #189: adding access roles endpoint to RDF lexicon (master...master) http://git.io/F5vrhQ
* github-ff leaves
<osmandin>afk 6012:14
* nbanks joins12:30
* travis-ci joins12:39
[travis-ci] futures/fcrepo4#1348 (master - b91bd10 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/0f3c594758ec...b91bd10e1de4
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15049057
* travis-ci leaves
* ermadmix leaves13:06
<cbeer>escowles: all your testing has been on infinispan-backed binary stores, right?13:10
<escowles>cbeer: yes -- i've only started working with projections today
<cbeer>escowles: ok. i'm actually looking at Modeshape's file binary store as a possible solution to my timeout problem. i guess i'll see what kind of performance i get out of that config too13:12
<barmintor>[INFO] There are 45 checkstyle errors.13:13
I'm horrible.
<awoods>barmintor: What is the threshold for "horrible"? 30 checkstyle errors?
<barmintor>awoods: I beleive that is a limit set by the tech lead13:14
<awoods>30 = horrible
osmandin: Let me know when you are back so we can discuss this: https://wiki.duraspace.org/display/FF/ModeShape+Artifacts+Layout13:15
* ermadmix joins
<awoods>If anyone else is interested in providing feedback on or dicussing the above linked wiki page, chime-in once osmandin returns.13:16
<pivotal-bot>Andrew Woods added comment: "Let me know when you have a moment to discuss, @osmandin." https://www.pivotaltracker.com/story/show/4901279913:17
Andrew Woods rejected "Introspect the bson output for modeshape schematics and report back how friendly the output is" https://www.pivotaltracker.com/story/show/49012799
<osmandin>awoods: on hangout?13:21
<awoods>osmandin: sure13:22
https://plus.google.com/hangouts/_/calendar/eW91cm1lZGlhc2hlbGYuY29tXzVlYzdpNXQ2Z282dTdidHI4aTVrbGJxOTUwQGdyb3VwLmNhbGVuZGFyLmdvb2dsZS5jb20.opn3ai06k1ge0no877ei3cku4o
* gregjansen leaves13:30
<pivotal-bot>Osman Din started "Introspect the bson output for modeshape schematics and report back how friendly the output is" https://www.pivotaltracker.com/story/show/4901279913:47
<escowles>a little poking showed the vast majority of time for viewing filesystem federated objects was loading the jcr:content node. i'm using the modeshape FileSystemConnector that's part of fcrepo4 -- should i be using a different version? fcrepo-filesystem-modeshape-federation-connector?13:49
* ermadmix leaves13:51
<awoods>escowles: Have you looked at this page from ermadmix: https://wiki.duraspace.org/display/FF/Design+-+Large+Files13:52
escowles: Specifically the section that details "New Modeshape FileSystemConnector option to compute hash from path rather than full content"13:53
<escowles>awoods: no, i was looking at fasseg 's large file wiki pages :-/
<awoods>escowles: I believe fasseg reported that navigating large files with the connector was very fast (with maybe a ~15sec delay on the first load)13:54
<escowles>awoods: yes, 15 sec for a 1GB file, but larger and larger delays for larger files (15+ mins for 8GB files)13:55
<awoods>escowles: It would be good to get ermadmix in the conversation.
* ermadmix joins
<awoods>speak of the ermadmix...
<escowles>ermadmix: we were just talking about you --13:56
i'm trying to serve large files with the filesystem connector, and awoods just pointed me at your wiki docs and the contentBasedSha1 option
i'm trying that out now
<awoods>it would be great to get your ideas about what kind of performance escowles should be able to expect, ermadmix.13:57
osmandin, can you get ermadmix's attention?
<osmandin>sure13:58
<ermadmix>escowles: I hear you...
escowles: with cotentBasedSha1=false hitting the fcr:content node will return instantaneously,13:59
<pivotal-bot>Osman Din added "Create simple utility that takes ModeShape filesystem binary artifacts (i.e. fcrepo4-data directory as input) and gener..." https://www.pivotaltracker.com/story/show/6207150414:01
Osman Din started "Create simple utility that takes ModeShape filesystem binary artifacts (i.e. fcrepo4-data directory as input) and gen..." https://www.pivotaltracker.com/story/show/62071504
<escowles>ermadmix: i've done 1, 2 and 4GB files and i'm seeing no delay for loading the object profile (page returning in less than 0.05 sec)
<ermadmix>...set to true - the node won't return until a sha1 is computed for the content of jcr:data, which can take ~ 60min for ~64GB file14:02
<escowles>oof -- glad i didn't wait that long
<pivotal-bot>Andrew Woods added comment: "Please review at your convenience, @escowles." https://www.pivotaltracker.com/story/show/6182355414:03
<barmintor>Performance tip: Try not to lookup nodes. In fact, try not to interact with the repository. At all. It's much faster.14:05
* gregjansen joins14:07
<barmintor>gregjansen: who is going to replace you on APT SG?14:12
<gregjansen>barmintor: SG?14:13
<barmintor>w/e that group you are on that was lamenting your pending absence yesterday
<gregjansen>barmintor: I do not know. The SG is taking nominations, recommendations
<barmintor>whaddayou gotta do so important you can't fix APT for us?14:14
<gregjansen>barmintor: I will remain on the tech group calls, but do not have time to be a tech liaison anymore
<mikeAtUVa>RDF party people: In what contexts (if any) can we have a reasonable expectation of meaningfully ordered triples? Does it even make sense to think about orders for triples?14:15
<escowles>mikeAtUVa: unless it's in a construct that's explicitly ordered (like rdf:List), you can assume it's unordered
<gregjansen>barmintor: UNC has plenty of software debt all it's own ;-)14:16
<mikeAtUVa>escowles: what about an iterator?
<barmintor>:(
<awoods>cbeer: Related to this comment on line 121, do you have a recommendation for a "smart" client that can be used to test FedoraBatch? https://github.com/futures/fcrepo4/pull/186/files#diff-748a2edaf2ead8aed31038150e2066d3R12114:18
<ajs6f>awoods: ping?
<awoods>ajs6f
<cbeer>awoods: anything that can construct a multipart/mixed request.14:19
<awoods>cbeer: suggestions?
cbeer: Or are you just speaking hypothetically?14:20
* github-ff joins
[fcrepo4] ajs6f created TurnOnHeaderJMS (+1 new commit): https://github.com/futures/fcrepo4/commit/507ec7a5f574
fcrepo4/TurnOnHeaderJMS 507ec7a ajs6f: Altered Spring config to use new JMS format
* github-ff leaves
* github-ff joins
[fcrepo4] ajs6f opened pull request #190: Altered Spring config to use new JMS format (master...TurnOnHeaderJMS) http://git.io/SoVNiQ
* github-ff leaves
<pivotal-bot>A. "Horbulaco" Soroka added "Turn on new JMS format in fcrepo-webapp" https://www.pivotaltracker.com/story/show/6207270014:21
A. "Horbulaco" Soroka started "Turn on new JMS format in fcrepo-webapp" https://www.pivotaltracker.com/story/show/62072700
A. "Horbulaco" Soroka finished "Turn on new JMS format in fcrepo-webapp" https://www.pivotaltracker.com/story/show/62072700
<ajs6f>awoods: ^^^
I forgot to turn on the new header-based message format14:22
When I wrote it.
It's a one-liner in the Spring config in fcrepo-webapp, and I don't know why we wouldn't just do it.
PR is waiting for you.
love, ajs6f
<escowles>mikeAtUVa: i'd still expect triples to be unordered, unless the underlying construct was ordered
<awoods>that is sweet, ajs6f. thanks
<pivotal-bot>Andrew Woods edited "Turn on new JMS format in fcrepo-webapp" https://www.pivotaltracker.com/story/show/6207270014:23
<ajs6f>mikeAtUVa: What they say. ^^^ If you want to talk about ordering, it's often a sign that the underlying model should be rethought.
<awoods>ajs6f: I will wait for travis-ci, then let it fly.14:24
<ajs6f>k
<cbeer>awoods: any client that lets you construct a multipart/mixed request (that doesn't assume every part is form-data) will do. probably means using a low-level MIME library in most places.14:25
(at least, that's how you construct multipart requests in ruby-land..)
<mikeAtUVa>escowles, ajs6f: it's just quite tempting to assume when that when you concat things to an RdfStream that you're making an assertion about order...14:26
<ajs6f>Evil does its job best when it is tempting.14:27
DON'T FALL FOR IT!
<escowles>yes, very understandable -- but RDF is very explicit about being unordered -- triples are all just independent statements and you can take them in any order, have to assume there are more statements out there, etc.
* mikeAtUVa only takes sips from the glass of RDF Kool Aid and often falls prey to the temptations from various serialization formats.14:28
<awoods>cbeer: I will poke around.
* escowles has been swimming in the RDF kool aid pool for years now14:29
<barmintor>mikeAtUVA: Have you heard the good news about resource identifiers and descriptive algebra?
<escowles>barmintor: i hope that's the setup for a really terrible pun..
<mikeAtUVa>barmintor: is that a new religion?
<barmintor>RDF IS YOUR GOD NOW14:30
<ajs6f>and OWL is the name of his prophet.14:31
<escowles>and blank nodes are the niggling inconsistencies that make you think the whole thing is a sham
<awoods>nice add, escowles... really drives it home.14:33
<ajs6f>escowles: Don't succumb to doubt. Try counting your DL rosary… AL, FL, EL, ALC, ALCOIN...
* escowles lights a stick of reification incense
* mikeAtUVa backs away slowly...
no seriously ... afk ~30m14:34
<ajs6f>Well, at least it isn't Scientoogy.
<cbeer>that's it. we don't need to store any data. we'll just reify it on access.
great ingest times. pretty lousy access.
* travis-ci joins
[travis-ci] futures/fcrepo4#1349 (TurnOnHeaderJMS - 507ec7a : ajs6f): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/commit/507ec7a5f574
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15055362
* travis-ci leaves
<barmintor>cbeer++ // my YK numbers wuld be awesome
<cbeer>or, fcrepo4: great for dark archives!
<ajs6f>If we can prove that we _could_ have retrieved it, why would anyone want to actually do it?
<barmintor>[INFO] There are 4 checkstyle errors.14:35
<awoods>5 < style bandit14:36
<ajs6f>0 will even pass the build for you.
awoods: Did we ever fix the 80-haracter thing?14:37
<pivotal-bot>Andrew Woods delivered "Turn on new JMS format in fcrepo-webapp" https://www.pivotaltracker.com/story/show/62072700
* github-ff joins
[fcrepo4] awoods deleted TurnOnHeaderJMS at 507ec7a: http://git.io/_d2idw
* github-ff leaves
<awoods>ajs6f: yes
<ajs6f>YAYYAYAYYYA!
<awoods>ajs6f: we fixed it to 120 characters.
<ajs6f>Works for me.
<awoods>ajs6f: I had to enable fcrepo-build-tools SNAPSHOTS to be available in our build, and to publish fcrepo-build-tools on all updates.14:39
<ajs6f>awoods: I'm glad you did.14:41
Why do we maintain our own build-tools atifact, anyway?
<awoods>ajs6f: build-tools just contains the style rules14:42
<ajs6f>awoods: Why wouldn't they be in fcrepo4?
<awoods>ajs6f: Which is suppose is handy for sharing across different repos.
<ajs6f>Hm. That's true
Urg. Solr is not the easiest thing to work with in a lightweight way for tests.14:43
<gregjansen>are we pulling jquery libraries at build time?14:47
I guess somebody is.. "com.gargoylesoftware.htmlunit.ScriptException: syntax error (http://ajax.googleapis.com/ajax/libs/jquery/1.9.1/jquery.min.js#12)"14:48
<ajs6f>We have integration tests that check for HTML goodness.
Maybe that's it?
<gregjansen>yeah that is it htmlunit
org.fcrepo.integration.http.api.html.FedoraHtmlResponsesIT14:49
<osmandin>benchtool fasseg: in the updated version of benchtool, is the default param -s bytes?14:51
* travis-ci joins14:52
[travis-ci] futures/fcrepo4#1351 (master - d4e8b10 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/b91bd10e1de4...d4e8b100ebf0
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15056355
* travis-ci leaves
<pivotal-bot>Eric James added comment: "needed to add org.fcrepo.transform to component-scan, will create PR" https://www.pivotaltracker.com/story/show/6194263815:08
* gregjansen leaves
* github-ff joins15:15
[fcrepo4] yulgit1 opened pull request #191: addded org.fcrepo.transform to rest.xml component scan (master...erj-transform) http://git.io/qtFHcQ
* github-ff leaves
<pivotal-bot>Eric James added comment: "https://github.com/futures/fcrepo4/pull/191" https://www.pivotaltracker.com/story/show/6194263815:16
<osmandin>awoods: just fyi, i re-ran benchtool on F3 and included the numbers: https://wiki.duraspace.org/display/FF/Test+-+Cluster+Setup+-+Yale15:17
<pivotal-bot>Eric James started "fcr:transform exception" https://www.pivotaltracker.com/story/show/6194263815:20
<ajs6f>afk bbs
* ajs6f1 joins15:32
* ajs6f leaves
* nbanks leaves15:33
* github-ff joins15:40
[fcrepo4] fasseg created fixity-projections (+1 new commit): http://git.io/Rnr0gA
fcrepo4/fixity-projections ef8d383 fasseg: added projection mixin type and a FileSystemConnector which is able to inject this mixin type
* github-ff leaves
* escowles leaves15:47
<awoods>thanks, osmandin16:03
<pivotal-bot>Eric James started "Documentation for fcrepo-transform's HTTP endpoints." https://www.pivotaltracker.com/story/show/6185044816:16
* ajs6f1 leaves16:19
* ajs6f joins16:23
Urg.16:25
https://issues.apache.org/jira/browse/JENA-576
This is annoying to the indexer because we're using Fuseki for testing.
Oh,well.
see y'all soon16:36
* ajs6f leaves
* barmintor just realized he has to re-run against master for apples-to-apples profiling16:39
Guh.
<awoods>cbeer: could you rebase PR-186 on master?
* github-ff joins16:43
[fcrepo4] cbeer force-pushed http-batch-refactor from 9035558 to 4026840: http://git.io/WjeANw
fcrepo4/http-batch-refactor 4026840 Chris Beer: Renamed FedoraDatastreams to generic FedoraBatch...
* github-ff leaves
<awoods>cbeer: thanks16:44
<cbeer>awoods: hold on. couple other problems with it.16:45
<awoods>cbeer: sure
cbeer: you will be happy to know that a triple-part POST works via curl, including: text, jpeg, and sparql-update
* github-ff joins16:48
[fcrepo4] cbeer force-pushed http-batch-refactor from 4026840 to 20d9c2a: http://git.io/WjeANw
fcrepo4/http-batch-refactor 20d9c2a Chris Beer: Renamed FedoraDatastreams to generic FedoraBatch...
* github-ff leaves
<cbeer>awoods: that should do it. i'm seeing some HTML IT failures, but I'd guess those are the ones that others have seen before.16:49
<awoods>cbeer: yes, nbanks and mikeAtUVa put in two updates to that test yesterday that fail intermittently.16:50
<mikeAtUVa>awoods: I can't seem to reproduce that one that fails consistently on jenkins.
<awoods>mikeAtUVa: hmm, I see it. and it sounds like cbeer may be as well.16:51
<mikeAtUVa>are other people seeing *that* error.
<cbeer>mikeAtUVa: which one? FedoraHtmlResponsesIT.testVersionCreationAndNavigation:209 There should be two revisions. expected:<2> but was:<0>
or one about namespaces..
<mikeAtUVa>cbeer: yeah, that's the one I can't reproduce. (expected 2 but got 0, or got 3 as jenkins reports)16:52
<awoods>mikeAtUVa: how are you testing, mikeAtUVa? just on that one test? or by running the full fcrepo-http-api project?16:53
<mikeAtUVa>awoods: mvn clean install
awoods: for everything
awoods: because I was woried that other data/settings might be interfering.
<awoods>hmm
<mikeAtUVa>don't get me wrong, there's still problems with my tests, but not relating to that assertion that I've discovered... except possibly some race codition relating to the HtmlUnit business.16:54
<awoods>mikeAtUVa: I wonder if we need to pass off the fix to someone who can reproduce it.
* osmandin leaves16:55
<mikeAtUVa>awoods: that would be fine... or if we want to @ignore that test (like we have others in that suite) I'd be ok with that too
awoods: I've sort of gotten sidetracked dealing with issues relating to an expectation that versions will be presented in chronological order on that page...16:56
<awoods>mikeAtUVa: Once you have done everything you can do, go ahead and @Ignore that test and create a bug ticket for bringing it back.16:57
cbeer: So far, the fcr:batch seems to work as advertised. However, I have not been able to set the mimeType on the EntityParts.17:02
<cbeer>awoods: because of client limitations?17:03
<awoods>cbeer: It is not clear. The debugger seems to be setting the correct mediaType on line 268 of FedoraBatch...17:04
cbeer: but the content-type is not apparent in the UI afterwards, and is not being reflected when I download datastreams.17:05
cbeer: I am posting my example to your wiki page. One moment.
<pivotal-bot>Mike Durbin added comment: "Versions aren't guaranteed to be in any order. I'll update the code to sort them by date to determine order." https://www.pivotaltracker.com/story/show/6201622017:09
* travis-ci joins17:10
[travis-ci] futures/fcrepo4#1356 (http-batch-refactor - 20d9c2a : Chris Beer): The build has errored.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/40268408f336...20d9c2a46760
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15063236
* travis-ci leaves
<awoods>cbeer: There it is: https://wiki.duraspace.org/display/FF/Design+-+Batch+Operations17:11
<cbeer>awoods: do you need a new line after the last part header? e.g. https://gist.github.com/cbeer/e805687ce03f6959c6ca17:16
(i don't know, just the difference i see between what the IT spits out and what you have there)17:17
<awoods>cbeer: the newline is there (just not in the wiki block for some reason)... and the three parts all work.17:18
<cbeer>ok. let me spin up fcrepo4 for real. i changed the IT to do something besides text/plain and it seemed like it worked17:19
<awoods>cbeer: you can download the real attachment and check that the size is correct: 2626, and run the command. It all works... just not the mimeType
<cbeer>awoods: i got a 404 on that.
<awoods>on what?17:20
<cbeer>the attachment.
<awoods>refresh
<cbeer>thanks.
* ermadmix leaves17:23
<cbeer>awoods: i'm seeing image/png set on ds1 here17:24
<awoods>correct
and the text on ds0
as well as the sparql on ds0
oh... you mean on download?
<cbeer>wait, what? i thought you said the content type wasn't being set correctly?17:25
<awoods>cbeer: where are you seeing image/png?
on the UI?
<cbeer><http://localhost:8080/rest/obs/ob0/ds1/fcr:content> <http://fedora.info/definitions/v4/repository#mimeType> "image/png"^^<http://www.w3.org/2001/XMLSchema#string> ;17:26
<awoods>cbeer: When I get the response via curl ( curl http://localhost:8080/rest/obs/ob0/ds1) , I see the triple you mention...17:27
cbeer: but I do not see it in the UI
<cbeer>...
<awoods>cbeer: and when I click on the content, it downloads instead of rendering the image in the browser.17:28
<cbeer>yes.
we're sending the Content-Disposition: attachment header on those downloads now
* github-ff joins17:29
[fcrepo4] mikedurbin opened pull request #192: Updated itests to make no assumptions about ordering in fcr:versions. (master...version-order-bug) http://git.io/GCti_g
* github-ff leaves
<awoods>cbeer: So it is not surprising... that triple not being rendered, and the content-type not being set on download?
<cbeer>and i see Content-Type:image/png
awoods: where aren't you seeing the triple? I see it under http://localhost:8080/rest/obs/ob0/ds1/fcr:content
<pivotal-bot>Mike Durbin added comment: "https://github.com/futures/fcrepo4/pull/192" https://www.pivotaltracker.com/story/show/62016220
<cbeer>fcrepo: mimeType
image/png
<pivotal-bot>Mike Durbin started "Write a velocity template for the fcr:versions endpoint." https://www.pivotaltracker.com/story/show/61474236
Mike Durbin added comment: "https://github.com/futures/fcrepo4/pull/192" https://www.pivotaltracker.com/story/show/61474236
<cbeer>awoods: and i'm seeing content-type set on download too.
<pivotal-bot>Mike Durbin finished "Write a velocity template for the fcr:versions endpoint." https://www.pivotaltracker.com/story/show/6147423617:30
<cbeer>unless chrome is making up headers
<awoods>cbeer: when I click on fcr:content... it downloads
<cbeer>curl "http://localhost:8080/rest/obs/ob0/ds1/fcr:content" -I
HTTP/1.1 200 OK
Content-Type: image/png
awoods: i guess i'm not sure what behavior you're seeing.17:31
<awoods>cbeer: ...It looks like we are golden. Sorry for the bother.
<cbeer>caching?
<awoods>cbeer: I was expecting different behavior from chrome.
<pivotal-bot>Mike Durbin added "Datastream versions created when content is updated should be easy to differentiate." https://www.pivotaltracker.com/story/show/6208512217:32
Mike Durbin edited "Datastream versions created when content is updated should be easy to differentiate." https://www.pivotaltracker.com/story/show/62085122
<mikeAtUVa>gotta go... have a nice weekend everyone....
* mikeAtUVa leaves
<awoods>bye...
I am stepping out for a moment, too, to hit the post office.17:33
<cbeer>awoods: we switched to sending the content disposition header so we could set e.g. the original file name, creation and modify dates, etc.
* github-ff joins17:41
[fcrepo4] barmintor created versionMgrReuse (+2 new commits): http://git.io/_e3RzA
fcrepo4/versionMgrReuse 5d53e2c Benjamin Armintor: allow VersionManager to reuse node to be versioned
fcrepo4/versionMgrReuse a30ef03 Benjamin Armintor: share sessions for change sets; publish FedoraEvents with info sufficient to gen JMS messages
* github-ff leaves
* github-ff joins
[fcrepo4] barmintor opened pull request #193: Version mgr reuse (master...versionMgrReuse) http://git.io/awWTdw
* github-ff leaves
<barmintor>oh, dammit
rebase later17:42
time for diner
* barmintor leaves
* ajs6f joins17:48
* ajs6f leaves17:51
* github-ff joins
[fcrepo-jms-indexer-pluggable] ajs6f created ImplMappingForSolrFields (+1 new commit): http://git.io/yrI7Cg
fcrepo-jms-indexer-pluggable/ImplMappingForSolrFields b7f2dde ajs6f: New mapping design for named-field indexers, not for RDF indexers, integration tests passing, no system tests yet
* github-ff leaves
* travis-ci joins17:59
[travis-ci] futures/fcrepo4#1359 (versionMgrReuse - a30ef03 : Benjamin Armintor): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/5d53e2cf63c9^...a30ef03eb8b3
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15066143
* travis-ci leaves
<pivotal-bot>Andrew Woods accepted "Turn on new JMS format in fcrepo-webapp" https://www.pivotaltracker.com/story/show/6207270018:33
* github-ff joins18:34
[fcrepo4] fasseg pushed 3 new commits to fixity-projections: http://git.io/9Ezh7Q
fcrepo4/fixity-projections 4509e3c fasseg: added CacheEntry interface and implementation for projected entries
fcrepo4/fixity-projections 2204187 fasseg: updated creation of fixity results so the new CacheEntry interface is used
fcrepo4/fixity-projections 31a45fa fasseg: updated fixity result model creation for projections and changed some checked exceptions from BinarystoreExcetions to the more genereal RepositoryException
* github-ff leaves
<pivotal-bot>Frank Asseg added comment: "Added a branch https://github.com/futures/fcrepo4/tree/fixity-projections" https://www.pivotaltracker.com/story/show/5600291618:35
* fasseg leaves18:45
* travis-ci joins18:50
[travis-ci] futures/fcrepo4#1360 (fixity-projections - 31a45fa : fasseg): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/ef8d383343ff...31a45facb25d
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15068388
* travis-ci leaves
* github-ff joins19:08
[fcrepo4] cbeer created html-content-properties (+1 new commit): http://git.io/u2B3CA
fcrepo4/html-content-properties 72b7e33 Chris Beer: Move content properties out of inlined resources on nt:file nodes
* github-ff leaves
* github-ff joins
[fcrepo4] cbeer opened pull request #194: Move content properties out of inlined resources on nt:file nodes (master...html-content-properties) http://git.io/ohhcow
* github-ff leaves
<pivotal-bot>Chris Beer added "Elevate content properties in the HTML display of nt:file nodes" https://www.pivotaltracker.com/story/show/6209001419:09
Chris Beer estimated "Elevate content properties in the HTML display of nt:file nodes" as 1 point https://www.pivotaltracker.com/story/show/62090014
Chris Beer started "Elevate content properties in the HTML display of nt:file nodes" https://www.pivotaltracker.com/story/show/62090014
Chris Beer finished "Elevate content properties in the HTML display of nt:file nodes" https://www.pivotaltracker.com/story/show/62090014
Chris Beer added comment: "https://github.com/futures/fcrepo4/pull/194" https://www.pivotaltracker.com/story/show/62090014
Andrew Woods edited "Fix up fcr:datastreams for a path-based world" https://www.pivotaltracker.com/story/show/6087699019:16
* travis-ci joins19:21
[travis-ci] futures/fcrepo4#1361 (html-content-properties - 72b7e33 : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/commit/72b7e33dcc26
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15069607
* travis-ci leaves
* github-ff joins19:27
[fcrepo4] awoods closed pull request #186: Renamed FedoraDatastreams to generic FedoraBatch (master...http-batch-refactor) http://git.io/ZtrJXA
* github-ff leaves
<pivotal-bot>Andrew Woods delivered "Fix up fcr:datastreams for a path-based world" https://www.pivotaltracker.com/story/show/6087699019:29
* ksclarke leaves19:30
* ksclarke joins19:31
* travis-ci joins19:40
[travis-ci] futures/fcrepo4#1363 (master - 20d9c2a : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/d4e8b100ebf0...20d9c2a46760
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15070166
* travis-ci leaves
<pivotal-bot>Andrew Woods delivered "Write a velocity template for the fcr:versions endpoint." https://www.pivotaltracker.com/story/show/61474236
* github-ff joins
[fcrepo4] awoods pushed 2 new commits to master: http://git.io/wOK_sQ
fcrepo4/master b9f07de Michael Durbin: Updated itests to make no assumptions about ordering in fcr:versions.
fcrepo4/master 5202a76 Andrew Woods: Merge pull request #192 from mikedurbin/version-order-bug...
* github-ff leaves
<pivotal-bot>Andrew Woods edited "Elevate content properties in the HTML display of nt:file nodes" https://www.pivotaltracker.com/story/show/6209001419:42
* travis-ci joins19:53
[travis-ci] futures/fcrepo4#1364 (master - 5202a76 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/20d9c2a46760...5202a76bab19
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15070704
* travis-ci leaves
<pivotal-bot>Andrew Woods delivered "Elevate content properties in the HTML display of nt:file nodes" https://www.pivotaltracker.com/story/show/6209001420:11
* github-ff joins
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/A213Mg
fcrepo4/master d64e7d3 Andrew Woods: Merge pull request #194 from futures/html-content-properties...
* github-ff leaves
<cbeer>awoods: does that help address our confusion from earlier?20:13
<pivotal-bot>Chris Beer added comment: "First round of docs added. Will add more next week with some more interesting examples." https://www.pivotaltracker.com/story/show/6087699020:15
* travis-ci joins20:23
[travis-ci] futures/fcrepo4#1365 (master - d64e7d3 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/5202a76bab19...d64e7d3cd085
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/15071817
* travis-ci leaves
<awoods>cbeer: what is that? the content-disposition comment?20:30
cbeer: I am heading out for dinner.20:31
cbeer: ...oh, you mean the mimeType being rendered. Yes, that is great.20:34
<pivotal-bot>Andrew Woods accepted "Write a velocity template for the fcr:versions endpoint." https://www.pivotaltracker.com/story/show/6147423622:41
Andrew Woods accepted "Fix up fcr:datastreams for a path-based world" https://www.pivotaltracker.com/story/show/60876990
Andrew Woods accepted "Elevate content properties in the HTML display of nt:file nodes" https://www.pivotaltracker.com/story/show/62090014