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

Using timezone: Eastern Standard Time
<pivotal-bot>Andrew Woods added "Fix kitchen-sink build" https://www.pivotaltracker.com/story/show/6187462400:54
Andrew Woods edited "Fix kitchen-sink build" https://www.pivotaltracker.com/story/show/61874624
* kaarefc joins
<pivotal-bot>Andrew Woods edited "Ensure HTML AJAX operations display all errors" https://www.pivotaltracker.com/story/show/6175345800:56
* kaarefc leaves01:14
* kaarefc joins
<pivotal-bot>Andrew Woods accepted "Single-node F3 Ingest Benchmarks" https://www.pivotaltracker.com/story/show/6055702401:21
Andrew Woods added comment: "Unit and integration tests required." https://www.pivotaltracker.com/story/show/6147423601:27
Andrew Woods rejected "Write a velocity template for the fcr:versions endpoint." https://www.pivotaltracker.com/story/show/61474236
Andrew Woods added comment: "@frankasseg?" https://www.pivotaltracker.com/story/show/6108774001:28
* ksclarke leaves01:33
<pivotal-bot>Andrew Woods accepted "Produce design doc for workflow messaging" https://www.pivotaltracker.com/story/show/6173229601:40
Andrew Woods accepted "Clean up bench-tool" https://www.pivotaltracker.com/story/show/6173501001:41
* kaarefc leaves01:42
<pivotal-bot>Andrew Woods added comment: "Please rebase this PR on master, as there are merge conflicts." https://www.pivotaltracker.com/story/show/6142169001:46
Andrew Woods rejected "Add alternate versioning policy." https://www.pivotaltracker.com/story/show/61421690
* github-ff joins02:02
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/Ah3vxA
fcrepo4/master 5213396 Andrew Woods: Merge pull request #179 from futures/content-disposition...
* github-ff leaves
<pivotal-bot>Andrew Woods delivered "Preserve uploaded file name as a node property (at least from the HTML UI; maybe more globally)" https://www.pivotaltracker.com/story/show/61500736
* kaarefc joins02:03
<pivotal-bot>Andrew Woods added comment: "Please rebase on master, as there are conflicts." https://www.pivotaltracker.com/story/show/6183102202:04
Andrew Woods rejected "Code formatting" https://www.pivotaltracker.com/story/show/61831022
Andrew Woods accepted "Preserve uploaded file name as a node property (at least from the HTML UI; maybe more globally)" https://www.pivotaltracker.com/story/show/6150073602:13
Andrew Woods edited "Code formatting" https://www.pivotaltracker.com/story/show/61831022
* travis-ci joins02:15
[travis-ci] futures/fcrepo4#1327 (master - 5213396 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/5aae3e199931...5213396c0e36
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14909308
* travis-ci leaves
* awoods leaves03:44
* awoods joins03:47
* jonathangee_ joins04:11
* jonathangee leaves04:12
* kaarefc leaves05:27
* kaarefc joins05:44
* kaarefc leaves05:59
* kaarefc joins06:07
* ajs6f joins08:38
barmintor: Saw your last comments in IRC— do you want to take it up with MODE (the missing events)?08:39
Might be a bug.
<awoods>ajs6f: I will push the following PR if you rebase it: https://www.pivotaltracker.com/story/show/6183102208:40
<pivotal-bot>feature: Code formatting (rejected) / owner: A. "Horbulaco" Soroka
<ajs6f>Funny, I thought that came straight off master. Oh, well. Will do.08:41
<awoods>ajs6f: I see the git commits yesterday are a little messed up... double commits.
<awoods>ajs6f: For example, these two have overlapping updates:08:45
<ajs6f>I have no idea what that's about. Those are not my work.08:46
Are you saying we have VC problems?
<awoods>ajs6f: I am just saying the merge conflict may be from a VC issue... or not.
ajs6f: Also, my master is not building, as of sometime yesterday.08:47
<ajs6f>awoods: No, I think the conflict is because escowles was working in the same code as was I: just a plain old stepping on toes deal.
<awoods>ajs6f: ok
ajs6f: Does the build work for you? I am failing in fcrepo-transform.
<ajs6f>Wait til I fix this merge, then I'll check.08:48
I don't want to juggle too much this early in the morning.
<ajs6f>You're only as old as you feel.
<pivotal-bot>A. "Horbulaco" Soroka added "Impl new design for indexing mappings" https://www.pivotaltracker.com/story/show/6189438808:52
* github-ff joins08:56
[fcrepo4] ajs6f force-pushed FormattingInFcrepoTransform from 3790f9c to c8ab96f: http://git.io/BEmnJg
fcrepo4/FormattingInFcrepoTransform c8ab96f ajs6f: Pure code reformatting
* github-ff leaves
<ajs6f>awoods: Now I'll try building master
<awoods>ajs6f: It looks like the fcrepo-transform build failure was mentioned yesterday: expecting ns001 instead of ns003.
ajs6f: And it seems that it is a platform-specific error.08:57
<ajs6f>awoods: escowles suspected it was his merge of stuff related to rdf:type triples.
<awoods>ajs6f: I ran into the same issue last week when I added an integration test that added a new namespace.08:58
ajs6f: It is probably the same test that is failing.
<ajs6f>awoods: That was your fault for adding tests. That's a nasty habit.
<awoods>ajs6f: The issue is that each new namespace gets an internal prefix (ns001, ns002, ..., ns00n)08:59
ajs6f: That test is doing a direct string comparison of the results...
<ajs6f>awoods: "Internal"? Internal to MODE? Internal to the RDF serialziation?
<awoods>ajs6f: I am not sure where the ns00X comes from, MODE or otherwise.09:00
ajs6f: I will gist the error
ajs6f: It would be best if we enabled some flexibility in the ordering of the namespaces in the tests, to expect a regex or something instead of a fixed string.09:01
<ajs6f>awoods: It would be correct to actually parse the returned material instead of relying on the particular characteristics of a serialization library. Then our tests would actually be testing what they claim to be testing.09:02
<awoods>ajs6f: even better, yes.
ajs6f: Do you know of a way to upload a file to gist? or is it all copy and paste?09:03
ajs6f: https://gist.github.com/awoods/778789209:04
<ajs6f>awoods: I've never tried.
(to upload)
awoods: I don't think that's really the intended purpose of Gist.
awoods: Master builds for me.09:05
(On a Mac)
<awoods>ajs6f: Not surprising.
<ajs6f>Did we check Java versions? Is it platform or JVM?
<awoods>ajs6f: But good to know
java -version
java version "1.7.0_09"
Java(TM) SE Runtime Environment (build 1.7.0_09-b05)
Java HotSpot(TM) 64-Bit Server VM (build 23.5-b02, mixed mode)
<ajs6f>d-137-155-234:fcrepo4 ajs6f$ java -version
java version "1.7.0_12-ea"
Java(TM) SE Runtime Environment (build 1.7.0_12-ea-b08)
Java HotSpot(TM) 64-Bit Server VM (build 24.0-b28, mixed mode)
The step actually makes a diff.09:06
I can run fcrepo-webapp with the version above.
But not with
<awoods>ajs6f: did you trace the cause?09:07
<ajs6f>(jdk1.7.0_40 is known to have problems on Mac OS 10.6, which I run.)
ajs6f: It looks like escowles mentioned the issue in his standup report email.09:08
<ajs6f>But he isn't likely to work on it today.
* travis-ci joins09:09
[travis-ci] futures/fcrepo4#1328 (FormattingInFcrepoTransform - c8ab96f : ajs6f): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/3790f9ca2ba7...c8ab96fa38b6
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14923416
* travis-ci leaves
<mikeAtUVa>ajs6f, awoods: two questions: 1 does master build for you guys? and to what list did escowles send this standup report e-mail?09:10
<ajs6f>awoods: I would merge ^^^ because it doesn't affect the JQLQueryVisitor thing one way or the other.
mikeAtUVa: Builds for me, not for awooods. See ^^^
<awoods>mikeAtUVa: I added you to the email list a few moments ago.
<ajs6f>mikeAtUVa: Are you on Linux?09:11
<mikeAtUVa>Doesn't build for me on linux. JQLConverterIT.testIt:206
<awoods>mikeAtUVa: same here
<ajs6f>Known problem.
escowles' fault.
I think he did it on purpose.
<awoods>mikeAtUVa: It is a string comparison issue related to the ordering of namespaces added during integration testing.
<mikeAtUVa>ajs6f: sadly the other two failures that cropped up when I rebased my code are on me;)
<ajs6f>mikeAtUVa: We get the rebases we deserve.09:12
<awoods>ajs6f: Did you say your PR is ready for merge?
<ajs6f>awoods: Yes.
awoods: It will fail for you with the same error, tho'.
<awoods>ajs6f: I will bring it down locally and give a spin (recognizing the known build error)09:13
<ajs6f>awoods: k
<ajs6f>No, not kk. Not that good.09:14
I'll let you know when it's kk.
<awoods>ajs6f: Do you have any interest in digging refactoring JQLConverterIT to do a more robust test?09:15
* kaarefc leaves09:16
<awoods>mikeAtUVa: I assume you have received the ff-internal emails?
<ajs6f>awoods: Sure, but I'm not in a good position to do anything immediately. Meetings until the afternoon, when I hope that ermadmix will have completed the refactor of indexing to accept the new message format (see my standup report).
awoods: But I would be happy to do it...09:17
<awoods>ajs6f: I saw it
<ajs6f>awoods: You may want to get cbeer or someone else on it to get it done quicker.
<awoods>ajs6f: maybe mikeAtUVa can see if anyone on the standup can get on it, otherwise tap on you.
<mikeAtUVa>awoods: I just got one, yeah. Thanks!
awoods, ajs6f: will do09:19
<awoods>mikeAtUVa: The IT failure is a simple one...
<ajs6f>mikeAtUva: You're a _standup_ guy. Get it? See what I did there?09:20
<awoods>ajs6f: booooo
<ajs6f>awoods: Criticism can be directed to my attorney's office.09:21
<awoods>mikeAtUVa: See this morning's IRC chat: http://irclogs.fcrepo.org/2013-12-04.html
* ermadmix joins
<awoods>ajs6f: Does ermadmix know you are waiting on his updates?09:22
<ajs6f>awoods: yep.
<awoods>ajs6f: That is very nice.09:23
<ajs6f>awoods: We are every nice people. We are so polite!
<awoods>mikeAtUVa: I reviewed two of your finished tickets. Please move them back into the "Finish"ed state to indicate you have addressed the issues.09:24
<pivotal-bot>A. "Horbulaco" Soroka started "Code formatting" https://www.pivotaltracker.com/story/show/61831022
A. "Horbulaco" Soroka finished "Code formatting" https://www.pivotaltracker.com/story/show/61831022
<mikeAtUVa>awoods: thanks... I will do. I've just got a few failing tests associated with the changes to the CND to address.
<awoods>mikeAtUVa: np... and by the way, I will be out for a few hours this morning, during standup.09:25
<mikeAtUVa>awoods: skiing?09:26
<ajs6f>awoods is on a call.
awoods is always on a call.
<awoods>mikeAtUVa: I call it "getting outside"
<ajs6f>awoods will always be on a call.
<mikeAtUVa>awoods: you work too hard, you should get outside as much as you can.09:27
<ajs6f>mikeAtUva: Really? I think of awoods more as sitting aroun eating peeled grapes while he laughs at PRs and throws merges into the trashcan.
* github-ff joins
[fcrepo4] awoods merged FormattingInFcrepoTransform into master: http://git.io/llalAA
* github-ff leaves
<ajs6f>That's what I do with my day.09:28
afk bbl
* ajs6f leaves
* ajs6f joins09:30
<pivotal-bot>Andrew Woods added comment: "Resolved with: https://github.com/futures/fcrepo4/compare/5213396c0e36...c8ab96fa38b6" https://www.pivotaltracker.com/story/show/61831022
Andrew Woods delivered "Code formatting" https://www.pivotaltracker.com/story/show/61831022
<awoods>mikeAtUVa: During standup, can you get the skinny on the status of the following two tickets:09:32
<pivotal-bot>feature: Federate over large files (rejected) / owner: Frank Asseg
feature: Improve benchtool performance: use pre-generated data, don't include object preparation in timings (delivered) / owner: Scott Prater
* kaarefc joins09:34
* nbanks joins09:36
<awoods>mikeAtUVa: Nevermind on: https://www.pivotaltracker.com/story/show/61568578 , it is closed.09:37
<pivotal-bot>feature: Improve benchtool performance: use pre-generated data, don't include object preparation in timings (delivered) / owner: Scott Prater
Andrew Woods added comment: "Overcome by: https://github.com/futures/benchtool/pull/7" https://www.pivotaltracker.com/story/show/6156857809:38
Andrew Woods accepted "Improve benchtool performance: use pre-generated data, don't include object preparation in timings" https://www.pivotaltracker.com/story/show/61568578
* travis-ci joins09:40
[travis-ci] futures/fcrepo4#1330 (master - c8ab96f : ajs6f): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/5213396c0e36...c8ab96fa38b6
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14924768
* travis-ci leaves
<pivotal-bot>Andrew Woods accepted "Code formatting" https://www.pivotaltracker.com/story/show/61831022
* kaarefc leaves09:43
<pivotal-bot>Nigel Banks added comment: "https://wiki.duraspace.org/display/FF/Converting+an+Fedora+3+Islandora+Content+model+to+Fedora+4+by+Example" https://www.pivotaltracker.com/story/show/6096348209:44
Nigel Banks finished "Write recipe for converting sample Fedora 3 content models to Fedora 4" https://www.pivotaltracker.com/story/show/60963482
* kaarefc joins
<pivotal-bot>Nigel Banks started "UI glossary documentation" https://www.pivotaltracker.com/story/show/61830274
<awoods>nbanks: There are at least 5 wiki pages related to content-modeling... maybe we can create a cohesive collection of pages? https://wiki.duraspace.org/label/FF/content-modeling09:46
<nbanks>awoods: Sure I'll collapse it into about three pages? In a tree structure?09:47
<pivotal-bot>Nigel Banks started "Write recipe for converting sample Fedora 3 content models to Fedora 4" https://www.pivotaltracker.com/story/show/6096348209:48
* kaarefc leaves09:49
<awoods>nbanks: That could be great... and it may be work touching base with Scott Prater (tecoripa) about moving those pages into the "Administrator Guide": https://wiki.duraspace.org/display/FF/Administrator+Guide
<pivotal-bot>Mike Durbin started "Add alternate versioning policy." https://www.pivotaltracker.com/story/show/61421690
<nbanks>awoods: Ok
<awoods>nbanks: The idea is that we start "graduating" wiki pages from the "Design - *" set (if they are ready) into the User and Administrator guides...09:50
* ksclarke joins
<awoods>nbanks: Likewise, we want to be doing a similar migration with the pages under FAQs towards the User and Administrator Guides.
<nbanks>awoods: Ok I'll gather all the pages under "Design - Content Modeling" then collapse them were appropriate. Then touch base with Scott before moving them under "User and Admin…"09:51
<awoods>nbanks: perfect
* ajs6f leaves09:52
* tecoripa joins09:54
<nbanks>awoods: "Fedora 3 Object representation in Fedora 4" already exists under another "Design -" Group perhaps I should leave it there? https://wiki.duraspace.org/display/FF/Design+-+Fedora+3+to+4+Upgrade09:56
<awoods>nbanks: You mean that page is duplicated?
nbanks: Or is it pulled into another page with the "include" macro?
<nbanks>awoods: Not duplicated, it only exists in the one place but it was also labelled as being "content model" related so it was in your search results.09:57
<awoods>nbanks: You can leave it unless you feel like the content belongs in the new set of pages... mikeAtUVa may have some input as well.09:58
nbanks, etal: I am stepping away for a few hours.
<nbanks>awoods: later
<pivotal-bot>Mike Durbin finished "Add alternate versioning policy." https://www.pivotaltracker.com/story/show/6142169009:59
<awoods>mikeAtUVa: I see your update... I will get it upon my return.
<mikeAtUVa>awoods: thanks, no rush.10:02
* github-ff joins10:03
[fcrepo-jms-indexer-pluggable] yulgit1 opened pull request #15: refactor IndexerGroup for new messaging format (master...erj-jmsheader1) http://git.io/yhGucQ
* github-ff leaves
<ermadmix>ajs6f: fcrepo-jms-indexer-pluggable PR 15 back to you10:06
* osmandin_ joins10:13
* gregjansen joins10:15
* osmandin_ leaves10:16
* osmandin joins10:18
<pivotal-bot>Frank Asseg added comment: "Modeshape 3.7 will fix this issue. This release will hopefully happen in Dec 13" https://www.pivotaltracker.com/story/show/6141839810:25
Frank Asseg started "fcr:fixity bug using the FileSystemConnector" https://www.pivotaltracker.com/story/show/56002916
* ksclarke leaves10:26
* ksclarke joins10:27
<tecoripa>cbeer: I'm doing some more investigation of the linux test problem in fcrepo-transform10:29
it seems to be a problem within the fcrepo-transform itself.10:30
here's what I've done:
mvn clean at the root (fcrepo)
* nbanks leaves10:31
<tecoripa>rmeove all fcrepo files in /tmp (the default datastore directory)
cd fcrepo-transform && MAVEN_OPTS="-Xmx1024m -XX:MaxPermSize=1024m" mvn clean install
and I get the same failure
namespace ns003 is created, instead of ns00110:32
<mikeAtUVa>tecoripa: yeah, the build is broken on linux.
<pivotal-bot>Frank Asseg added comment: "@ericjames: right.. it seems the fixity service somehow checks the FileCacheStore rather than the projection..." https://www.pivotaltracker.com/story/show/56002916
<tecoripa>mikeAtUVa: yes, we discovered that last night -- I'm seeing if I can narrow the problem down a bit.10:33
<mikeAtUVa>tecoripa: ajs6f and awoods discussed a fix this morning and we'll see who wants to fix it at the standup.10:34
<tecoripa>mikeAtUVa: okay, I'll wait to hear back, then. thanks
<mikeAtUVa>tecoripa: the info is in the logs... http://irclogs.fcrepo.org/2013-12-04.html, but I should make a tracker ticket with the details...
* nbanks joins10:37
<pivotal-bot>Mike Durbin added "JQLConverterIT.testIt fails at line 206 on linux." https://www.pivotaltracker.com/story/show/6190235810:38
Mike Durbin edited "JQLConverterIT.testIt fails at line 206 on linux." https://www.pivotaltracker.com/story/show/61902358
Frank Asseg added comment: "Problem is there is no filename/fixity/<nr> hierarchy available for the Nodeservice in the projected filesys..." https://www.pivotaltracker.com/story/show/5600291610:42
<tecoripa>nbanks: Your Islandora content/model cnd conversion documentation is good. I'm going to crib some of it to document a more general approach.10:48
<nbanks>tecoripa: thanks!10:49
<pivotal-bot>Eric James added comment: "Yes in DatastreamService the getFixity() looks in the LLeevelStorageService. So perhaps how to determine tha..." https://www.pivotaltracker.com/story/show/5600291610:50
<mikeAtUVa>standup https://plus.google.com/hangouts/_/calendar/eW91cm1lZGlhc2hlbGYuY29tXzVlYzdpNXQ2Z282dTdidHI4aTVrbGJxOTUwQGdyb3VwLmNhbGVuZGFyLmdvb2dsZS5jb20.opn3ai06k1ge0no877ei3cku4o11:00
cbeer, barmintor, ermadmix, techoripa, coming?11:01
* ajs6f joins11:05
<ajs6f>My standup: I'll review and merge ermadmix's PR for the indexer gear, and then move on to https://www.pivotaltracker.com/story/show/6189438811:10
<pivotal-bot>feature: Impl new design for indexing mappings (unscheduled) / owner: A. "Horbulaco" Soroka
<cbeer>huh. so it fails on jenkins, but not on travis.11:17
<ajs6f>cbeer: Travis isn't running the ITs.
<cbeer>ajs6f: oh. takes too long?
<ajs6f>cbeer: IDK, but that seems plausible.
<cbeer>mikeAtUVa: i'm looking at it now. that test really doesn't make any sense to me right now11:22
oh. at least i didn't write it.11:23
<mikeAtUVa>cbeer: In that it can be refactored to make sense, or that it doesn't make enough sense that you could recreate it in a working way?
<cbeer>mikeAtUVa: it looks like awoods actually saw this problem before, maybe.
he checks if it's ns001 or ns002
but not ns003.
<pivotal-bot>Chris Beer started "JQLConverterIT.testIt fails at line 206 on linux." https://www.pivotaltracker.com/story/show/6190235811:28
<ajs6f>We ought to be testing for the behavior we want to guarantee, which (as I understand it) is the presence of the namespaces, not the particular prefixes which which they were serialized.
<mikeAtUVa>ajs6f: thanks for clarifying to cbeer where we want to go...11:30
<ermadmix>ajs6f: just FYI,The InderGroupIT test class has wait loops w/o timeouts, you might want to add one, I had a process running all night.
<ajs6f>mikeAtUVa: I'm just spewing words here. It's what I do.
ermadmix: Yes, I noticed that ysterday and commented to escowles about it.11:31
ermadmix: I can do later this afternoon.
<cbeer>ajs6f: i thought i'd just loop from 0 to INT_MAX and hope one of them matches.. you're saying i shouldn't do that?
<ajs6f>ermadmix: I put an example in fcrepo-jms to copy.
<pivotal-bot>Nigel Banks edited "Write recipe for converting sample Fedora 3 content models to Fedora 4" https://www.pivotaltracker.com/story/show/60963482
<ajs6f>cbeer: I think we should never, ever, be testing serialized results in an IT. We should be testing a parsed form.11:32
cbeer: Because if the results don't parse, the test (no matter what it is testing) should fail.
<pivotal-bot>Chris Beer added comment: "https://github.com/futures/fcrepo4/commit/9447488014585d512b6ace668d17e17504add8e0" https://www.pivotaltracker.com/story/show/6190235811:33
<cbeer>ajs6f: yeah, i'm going to let you figure that out.11:34
<ajs6f>cbeer: It's what I would do in that IT that's failing. Parse, then test the resulting object for the presence of the wanted namesapce.
cbeer: It will never fail in this annoying way again.
cbeer: I _promise_.11:35
<cbeer>ajs6f: have you looked at the test?
* barmintor loves it when he has to look at MODE source
<pivotal-bot>Chris Beer finished "JQLConverterIT.testIt fails at line 206 on linux." https://www.pivotaltracker.com/story/show/61902358
<ajs6f>cbeer: Yeah, but I couldn't make heads or tails of it. But I didn't spend much time with it, because it only fails on Linux, which we don't support.
* github-ff joins
[fcrepo4] cbeer pushed 1 new commit to master: http://git.io/bCGepQ
fcrepo4/master 9447488 Chris Beer: pull the registered namespace prefix out of JCR before testing the query with it
* github-ff leaves
<cbeer>ajs6f: ok. i can't understand what you're asking in the context of that test, but that ^ ought to fix it.11:36
<ajs6f>cbeer: I'm not too worried about it. I just want awoods to stop complaining because his obscure OS has problems with our perfectly good code.11:37
<mikeAtUVa>cbeer: the test passes...11:38
<mikeAtUVa>ajs6f: them's fightin' words
cbeer: thanks a bunch!
<ajs6f>Come on, tecorpia/mikeAtUVa: step up to the beauty of a microkernel architecture. Use Plan 9.11:39
<tecoripa>I hear that ajs6f still uses Windows 3.1 at home.
<ajs6f>tecoripa: CP/M.
<barmintor>I've also heard that
You can actually hear the Win3.1 startup chime during standup, if you listen closely.11:40
Usually a few times.
<tecoripa>wasn't that in a Dickens Christmas story? the ghost of OS past?11:41
<barmintor>"Why aren't those children playing, Ghost?" "They're waiting on the Microsoft helpline, Ebeneezer." "Why are they dressed so poorly?" "They're charged $15 if the problem is demonstrated to be user error, Ebeneezer."11:43
"That's barbaric! Why don't they just google their problem?"11:44
"The help pages are behind a paywall, Ebeneezer."
<ajs6f>"Bing", barmintor. The verb is "Bing".
<barmintor>"bing" is not available to the Ghost of OS-Past11:45
"altavista" is acceptable
<pivotal-bot>Chris Beer delivered "JQLConverterIT.testIt fails at line 206 on linux." https://www.pivotaltracker.com/story/show/61902358
Chris Beer accepted "JQLConverterIT.testIt fails at line 206 on linux." https://www.pivotaltracker.com/story/show/61902358
Chris Beer edited "Make property values in HTML UI editable" https://www.pivotaltracker.com/story/show/6078128011:47
* travis-ci joins11:55
[travis-ci] futures/fcrepo4#1332 (master - 9447488 : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/c8ab96fa38b6...944748801458
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14931796
* travis-ci leaves
* ajs6f leaves11:56
<tecoripa>build succeeded. ajs6f can go back to playing Jazz Jackrabbit (http://en.wikipedia.org/wiki/Jazz_Jackrabbit_%281994_video_game%29)12:00
cbeer, mikeAtUVa: thanks!12:01
<barmintor>crap. I think this is a bug in MODE
<tecoripa>mikeAtUVa: can you tell me that status of versioning in this latest build? is it enabled or disabled by default now?12:02
<cbeer>barmintor: what do you mean, crap? i think you mean "yay, someone else has to fix it!"
<barmintor>cbeer: but I have to demonstrate it
* ajs6f joins12:04
<bljenkins>Yippie, build fixed!12:05
Project fcrepo4 build #1488: FIXED in 29 min: http://ci.fcrepo.org/jenkins/job/fcrepo4/1488/
Chris Beer: pull the registered namespace prefix out of JCR before testing the query with it
<ajs6f>ermadmix: Have you done: https://www.pivotaltracker.com/story/show/61750308?12:06
<pivotal-bot>feature: jms-indexer-pluggable - comply to header messaging (started) / owner: Eric James
<bljenkins>Project fcrepo-fixity-corrupter build #524: SUCCESS in 1 min 59 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/524/12:07
<pivotal-bot>Scott Prater edited "Write recipe for converting sample Fedora 3 content models to Fedora 4 Node Types" https://www.pivotaltracker.com/story/show/60963482
<bljenkins>Project fcrepo-kitchen-sink build #694: FAILURE in 2 min 20 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/694/12:09
<cbeer>ajs6f: i'm trying to figure out what to do with https://www.pivotaltracker.com/story/show/6087699012:11
<pivotal-bot>feature: Fix up fcr:datastreams for a path-based world (unstarted) / owner: Chris Beer
<cbeer>ajs6f: i kinda want to turn it into fcr:batch
<ajs6f>There's no description for that ticket.
<ermadmix>ajs6f: the first task of that ticket is basically that PR I submitted, I thought you wanted to do the other 2 so I held off on them. Let me know.
<ajs6f>ermadmix: np, somehow I missed the PR. I'll go review and merge now.12:12
<cbeer>ajs6f: and let you post some combination of triples and binary content, somehow with appropriate content dispostions.
<ajs6f>cbeer: Like Chris Wilper's old Semantic Packages idea?
<cbeer>ajs6f: maybe i should wiki this, actually, and then get your feedback.12:13
ajs6f: let me look at cwilper's stuff.
aww. http://www.ietf.org/id/draft-wilper-semantic-content-pkgs-00.txt is 404
ajs6f: but, yeah.. multipart/mixed could be one expression at fcr:batch.. and someone could do OAI-ORE or whatever else there too12:14
<ajs6f>cbeer: https://github.com/cwilper/scpproject.org
cbeer: It's an intriguing thought— how much does it overlap with our already-impl'd transactions?
<cbeer>ajs6f: one HTTP request.
<ajs6f>cbeer: right. is that a big use case? It's an honest question… I really don't know.12:15
<cbeer>hm. i don't want to do SCP. i'd rather lean on HTTP machinery to start.
ajs6f: it is a big use case to making fcrepo4 fast.
<ajs6f>cbeer: Really? That surprises me, but I believe you. If that's so, then it's worth it, in my mind.
<cbeer>(as long as we forget eddies ever did batch stuff for fcrepo3)12:16
<ajs6f>Who is eddies?
cbeer: So, multipart with graph and bitstreams?
<barmintor>cbeer++ // ajs6f: the client overhead in making a series of calls is significant
<cbeer>ajs6f: wasn't he the barista at the fcrepo4 houses?12:17
<ajs6f>cbeer/barmintor: I believed cbeer, but now that barmintor agrees, I have to dissent.
<ajs6f>cbeer: I think he was the guy who drank so much.
<cbeer>ajs6f: yeah, let me wiki this and see where it goes from here.12:18
<ajs6f>cbeer: k. It sounds like a good start.
<cbeer>ooh. i might go overboard with this content-disposition-as-a-request-header stuff.12:19
<ajs6f>cbeer: Let's not commit to this HTTP thing until we see whether it has any legs.12:20
<cbeer>we could use it in POST /path/to/node to distinguish between triples for the object and triples for a binary datastream
<ajs6f>ermadmix: Getting compilation errors from that PR- see my comment there.12:24
<ermadmix>ajs6f: sorry, added that line w/o testing, fixing now12:25
<nbanks>is registering namespaces working for anyone else in the HTML interface? I've tried it from the command line as well but I don't see my namespaces in the HTML interface.12:26
<ajs6f>ermadmix: np, I should be so lucky as to have so few problems with my PRs.
<ermadmix>ajs6f: fixed12:28
<nbanks>Hmm seems it's doing something like:12:38
    <http://islandora.ca/ri> <http://fedora.info/definitions/v4/repository#hasNamespace> "prefix"
} WHERE { }
Where as the integration tests are doing something like this:
    <uri> <http://purl.org/vocab/vann/preferredNamespacePrefix> "prefix"12:39
} WHERE { }
<cbeer>ajs6f: https://wiki.duraspace.org/display/FF/Design+-+Batch+Operations12:42
<nbanks>experimenting makes it look like "http://purl.org/vocab/vann/preferredNamespacePrefix" is the correct predicate, I'm gonna make a ticket.12:43
<cbeer>nbanks: want to fix that? and write an HTML IT for it?12:44
<nbanks>cbeer: Sure will do
<cbeer>nbanks: there's some other JS tests in FedoraHtmlResponsesIT
<nbanks>cbeer: I'll stick the test in there then
<pivotal-bot>Nigel Banks added "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/6191498012:46
Nigel Banks started "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/61914980
<ajs6f>cbeer: I'm confused by the semantics of name="". Is it a path? Is it a verb? is it a plane? Is it Turkish Superman?12:47
<cbeer>ajs6f: node path relative to whereever you posted.12:48
i assume the modeshape path factory will do all the heavy lifting there.
<ajs6f>cbeer Then what about "delete[]"?
<cbeer>ajs6f: special semantics.
<cbeer>ajs6f: that replaces the query parameter that we're currently using.
ajs6f: and it's marked up as form-data, so i think we can get away wtih it12:49
<ajs6f>On the batch endpoint?
<cbeer>ajs6f: yup. ugly hack alert: "http://localhost:8080/rest/path/to/some/resource/fcr:datastreams?delete=old-content"
Well, the whole deal with packaging operations over HTTP12:50
is that you can't do normal HTTP.
<osmandin>afk 30
<ajs6f>cbeer: Can we define a new parm on Content-Disposition for verb/12:51
Is tht koshe?
Is that kosher?
Stupid keyboard.12:52
<cbeer>ajs6f: pretty sure they leave it totally open.
<ajs6f>cbeer: I'd like that better than overloading name.
<cbeer>http://www.w3.org/Protocols/rfc2616/rfc2616-sec19.html 19.5.1
disposition-type = "attachment" | disp-extension-token12:53
not sure what token allows though.
<ajs6f>For us it could just be [update|delete].
Those are the only two verbs, right?12:54
Updating a non-existent thing is ingest.
<cbeer>ajs6f: what's wrong with inline + application/sparql-update for doing updates? at least it's kinda standard, and it's unambiguous for us, at least.12:55
token = 1*<any CHAR except CTLs or separators>
<ajs6f>cbeer: Fine, fine. I just want to get it out of name, so that name's semantic is totally simple.
Get delete out of name, that is.12:56
* tecoripa leaves
<cbeer>ajs6f: sure. it makes it easier to document, at least (although, arguably, those names are server-defined opaque strings anyway.)12:57
we just happen to do interesting things with attachments and inline values
<ajs6f>cbeer: k. You know how neurotic I am about these things.
I'm going to go do some interesting things with bean tacos. Like eat 'em.12:58
<cbeer>ok. shipping it. maybe without deletes to start. it's not like those are expensive deletes anyway.
<barmintor>Aargh. cbeer, did you have to do any weird magic to get MODE imported into eclipse?13:04
<cbeer>barmintor: very weird magic. used intellij :P
* ermadmix leaves
* osmandin leaves13:08
<cbeer>barmintor: does this have anything to do with MODE's special settings.xml, maybe?13:18
mikeAtUVa: was it you or gregjansen that had some way to get at the path value factory?13:19
<barmintor>cbeer: I assume so, but those profiles are already in my settings.xml
<mikeAtUVa>cbeer: likely gregjansen since that doesn't ring a bell to me.
* ermadmix joins13:24
* tecoripa joins13:27
* github-ff joins13:45
[fcrepo-jms-indexer-pluggable] ajs6f created pr/15 (+3 new commits): http://git.io/Z9CSgg
fcrepo-jms-indexer-pluggable/pr/15 34d78db Eric James: refactor IndexerGroup for new messaging format
fcrepo-jms-indexer-pluggable/pr/15 85b72f5 Eric James: fixed bug
fcrepo-jms-indexer-pluggable/pr/15 02c6f4a ajs6f: Adding timeouts to tests, stronger Sparql indexer
* github-ff leaves
* github-ff joins13:47
[fcrepo-jms-indexer-pluggable] ajs6f force-pushed pr/15 from 02c6f4a to c38a9e7: http://git.io/Niir7g
fcrepo-jms-indexer-pluggable/pr/15 c38a9e7 Eric James: refactor IndexerGroup for new messaging format, adding timeouts to tests, stronger Sparql indexer
* github-ff leaves
* github-ff joins13:48
[fcrepo-jms-indexer-pluggable] ajs6f opened pull request #16: refactor IndexerGroup for new messaging format, adding timeouts to tests... (master...pr/15) http://git.io/e8otMw
* github-ff leaves
* github-ff joins
[fcrepo-jms-indexer-pluggable] ajs6f closed pull request #15: refactor IndexerGroup for new messaging format (master...erj-jmsheader1) http://git.io/yhGucQ
* github-ff leaves
* github-ff joins13:49
[fcrepo-jms-indexer-pluggable] ajs6f pushed 1 new commit to master: http://git.io/PeBiOg
fcrepo-jms-indexer-pluggable/master 4a41346 ajs6f: Merge pull request #16 from futures/pr/15...
* github-ff leaves
<tecoripa>question about content models reborn as node types:
should the user-defined cmodel node type be a child of fedora:resource, or fedora:object?
<pivotal-bot>A. "Horbulaco" Soroka added comment: "Resolved with https://github.com/futures/fcrepo-jms-indexer-pluggable/commit/4a41346d8e98ba14a62cd..." https://www.pivotaltracker.com/story/show/61750308
A. "Horbulaco" Soroka finished "jms-indexer-pluggable - comply to header messaging" https://www.pivotaltracker.com/story/show/6175030813:50
A. "Horbulaco" Soroka delivered "jms-indexer-pluggable - comply to header messaging" https://www.pivotaltracker.com/story/show/61750308
<tecoripa>looking at the fedora-node-types.cnd, I'm inclined user-defined types should be children of fedora:resource, since fedora:object just seems to be a way to say "this is an object".
but I could be convinced otherwise.
<gregjansen>hey guys, I will be on a call for the rest of today.. :-(13:55
* ermadmix leaves13:56
<cbeer>ajs6f: ok, i've figured out why i didn't do this whole multipart thing before... the apache commons mimetype API is awful.13:57
so, short of building up requests by hand, i guess, it's almost untestable.
or i haven't found the right thing to do, because they don't have any javadocs.13:58
<ajs6f>cbeer: Why not use the really good other HTTP client?14:00
<cbeer>ajs6f: the one in jax-rs 2?
<ajs6f>cbeer: Sure. I wasn't thinking of any particular one, because I don't know of any. I was just assuming there must be one out there.14:01
That's how I live my life.
* ermadmix joins14:02
* tecoripa leaves14:43
The Sparql-based triplestore "indexer"...
it does removes by running a DESCRIBE against the triple store for a given URI (pid).14:44
Then it deletes all of those triples.
But we're producing lots of triples (e.g. for versions) that wouldn't be caught be that.
Oh, well. I'm only interested in Solr.
<cbeer>jersey-- # "The content dispostion type is not equal to form-data"15:06
<pivotal-bot>Andrew Woods added comment: "That is good news. It will not be in time for the holiday release, but will be a welcome fix shortly therea..." https://www.pivotaltracker.com/story/show/6141839815:09
Chris Beer added comment: "Isn't this worth waiting for? " https://www.pivotaltracker.com/story/show/6141839815:10
<barmintor>oh, for Pete's sake15:16
that's not a bug, that's the specified behavior
"Only events whose associated parent node has one of the node types (or a subtype of one of the node types) in this list will be received."
That is great for Properties, but totally pointless for Node events
<nbanks>cbeer: Man those integration tests drove me nuts, turns out you have to wait for the page via a separate function call before you can see it's plain text. I just assumed the html on the page would be rendered by that point.15:23
* ajs6f leaves
* ajs6f1 joins
* github-ff joins15:25
[fcrepo4] nigelgbanks opened pull request #185: Add Namespace in browser now functioning, also added an integration test (master...61914980-namespace-prefix) http://git.io/MTkHsw
* github-ff leaves
<pivotal-bot>Nigel Banks added comment: "https://github.com/futures/fcrepo4/pull/185" https://www.pivotaltracker.com/story/show/6191498015:26
Nigel Banks finished "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/61914980
Nigel Banks edited "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/6191498015:28
Nigel Banks edited "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/61914980
<barmintor>that spec is broken15:33
I can't see how nodeType is supposed to be the context node's type for properties, but the parent's node tpye for nodes
but whatever
I'll build around jcr:created changes15:34
assuming they publish property removal events when a node is removed
<pivotal-bot>Chris Beer started "Fix up fcr:datastreams for a path-based world" https://www.pivotaltracker.com/story/show/6087699015:35
<ajs6f1>barmintor: Eh? You're talking about the Observation stuff?
<barmintor>ajs6f1: yes15:36
<ajs6f1>barmintor: If you're sure the spec is broken, you want me to get into it with MODE or the JCR group?
{puts on boxing gloves}
<barmintor>ajs6f1: You tell me if it seems broken to you
<ajs6f1>barmintor: I haven't been into it like you. Where am I looking?15:37
<barmintor>if you listen to, say, events for nt:file
to me, this means "I want events when the Node context has type nt:file"
<pivotal-bot>Andrew Woods added comment: "I much prefer sticking to a release date, and rolling improvements in two to four weeks later. I do not wan..." https://www.pivotaltracker.com/story/show/61418398
<ajs6f1>Makes sense.
<barmintor>but what it is specified to mean is:15:38
"I want Property events when the node context has type nt:file, but I want Node events when the *parent* of node context is nt:file"
<cbeer>barmintor: i guess that make sense.. jcr doesn't really treat properties and node-children that different
<ajs6f1>I'm trying to think about that. I don't get it. It seems bad-surprising.15:40
<barmintor>cbeer: yeah, but this means you cannot simply register a listener for "node added of type nt:file"
<ajs6f1>barmintor: Isn't there a param for "listen to events from spaces under this node"?
<cbeer>barmintor: right. maybe we could get that in as a modeshape feature?
<ajs6f1>barmintor: Could that have something to do with it?
<barmintor>ajs6f1: there is, but it seems unlikely that the intended practice is to put all nodes of a type under a ommon parent15:41
<ajs6f1>barmintor: Np, that wouldn't make any sense. I'm just wondering about the fact that it looks at the _parent_. IDK. It's definitely not what I would have expected.15:43
<pivotal-bot>Chris Beer added comment: "It just seems really weird to say "hey, here's a release you all can use.. but don't use big files yet". We'r..." https://www.pivotaltracker.com/story/show/6141839815:45
Andrew Woods added comment: "The 13th came from @frankasseg's comment above." https://www.pivotaltracker.com/story/show/6141839815:47
<gregjansen>cbeer: still looking for a path value factory?15:49
<cbeer>gregjansen: if you have anything better than making a new ExecutionContext, i'll take it15:50
<gregjansen>cbeer: okay, looking
<cbeer>final PathFactory pathFactory = new ExecutionContext().getValueFactories().getPathFactory();15:51
^ that
<gregjansen>cbeer: I'm not even sure that path factory would have everything it needs. I recall not finding a way to construct it.15:53
<cbeer>gregjansen: yeah. to construct it you need the valueFactories
<gregjansen>cbeer: looks like I avoided it in the long run15:54
cbeer: which is a much bigger thing..
<pivotal-bot>Andrew Woods added comment: "@cbeer, what does your availability look like between Dec 13 and Dec 24? If you can commit to being around ..." https://www.pivotaltracker.com/story/show/6141839815:58
<cbeer>awoods: "As it stands right now, the F4 dev team scatters on Dec 13".. is it so much that we scatter, as we're just off the project?15:59
<gregjansen>mikeAtUva: question, is the current linux issue an IllegalStateException from spring's WebApplicationContextUtils?16:07
<nbanks>For the HTTP API Do we allow the user to create new objects with user defined mixins? I can't seem to do it, or modify the fcrepo:mixinTypes property after a normal fedora:object is created.16:08
<mikeAtUVa>gregjansen: current linux issue? Nope, Chris fixed the bug in JQLConverterIT... I'm unaware of any issue in WebApplicationContextUtils.16:09
<nbanks>I get why we wouldn't allow the mixinTypes to change after creation, but we should be able to specify custom mixins at ingest?
<awoods>cbeer: I am not sure what the F4 dev team is doing after Dec 13. All I know is that there are no commitments of anyone being around for F4 work.16:17
nbanks: I assume you are looking at: https://wiki.duraspace.org/display/FF/HTTP+API#HTTPAPI-Objects (POST)?16:18
* ajs6f1 leaves16:21
* tecoripa joins
<cbeer>awoods: ok. if that's all the problem is, let me check with tcramer16:22
<nbanks>awoods: yup16:23
* ajs6f joins
<awoods>cbeer: As long as we get the release out before Christmas... we are fine.
* ajs6f leaves
<awoods>nbanks: And the POST object call documented in the HTTP wiki is not working for you?16:26
<nbanks>awoods: It works as document in that you can create objects/datastreams but it seems as though I can't create any other registered types that extend the fedora:object type.16:27
awoods: s/document/documented16:28
<awoods>nbanks: I thought mikeAtUVa successfully added a mixin to an existing object just this week. mikeAtUVa?16:29
<mikeAtUVa>nbanks, awoods: yeah, I added it after object creation though... https://github.com/mikedurbin/fcrepo4/commit/fab2d9e41f9495ee670cfb7447ed702aea4a6751#diff-2e25ac74722dfade60d390a37f688725R22216:32
nbanks: INSERT DATA { <> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <mixinurl> . }16:33
<nbanks>mikeAtUVA: Ok I'll give that a go, any thoughts as to opening up the HTTP API so we can create objects that extend fedora:object16:34
<mikeAtUVa>nbanks: while I'm not opposed to such an idea, I'd be interested to see where the use of mixins is insufficient or less useful in accomplishing the goals of content modeling...16:36
nbanks: or if you just mean adding a mixin upon creation... that's certainly useful.. as would be setting properties as part of atomic creation... though maybe the import function was meant for this?16:37
<nbanks>mikeAtUVA: my concern would be for application logic, if I can be reasonably search that the object I created had all the mandatory properties and children for later processing
<tecoripa>nbanks: I've been thinking that our cmodel node types would be sub-types of fedora:resource
<nbanks>mikeAtUVa: hmm ya import makes sense, I didn't think of that.
<tecoripa>since fedora:object is used to say that a node is an object, nothing more
<nbanks>mikeAtUVa: /s/search/sure16:38
<tecoripa>nbanks, mikeAtUVa: not sure if this is relevant, but cbeer and I had a conversation about this last week
<nbanks>I may also be doing something wrong in that I have my type islandora:bookCModel extending fedora:object like so
[islandora:bookCModel] > fedora:object orderable mixin
<tecoripa>if your node type contains some manadatory properties or child nodes, you need to add the properties and children to the object *before* you add the type16:39
<nbanks>tecoripa: Please share.
<tecoripa>or else all heck breaks loose.
<awoods>nbanks/cbeer: As it stands right now, the POST object?mixin=<mixin-type> is limited to which mixin-types? fedora:datastream and fedora:object?16:40
* github-ff joins
[fcrepo4] cbeer created http-batch-refactor (+1 new commit): http://git.io/F35Pww
fcrepo4/http-batch-refactor 9035558 Chris Beer: Renamed FedoraDatastreams to generic FedoraBatch...
* github-ff leaves
<nbanks>awoods: I think so
<tecoripa>cbeer thought this was an area that needed some more testing and work, as object validation (in terms of the structure of the object, no the content of the nodes) was highly dependent on the node types associated with it
<nbanks>tecoripa: In all hell breaks loose do you mean it creates a bunch of random child objects, cause thats what I'm seeing although it looks like a bug in the HTML interface.
<mikeAtUVa>nbanks, tecoripa: that would seem to push us towards a model where everything is ingested as a fedora:object and content models (which could still extend fedora:object) are added as needed.
<tecoripa>and could cause lots of pain in an object creation/modification workflow
* github-ff joins16:41
[fcrepo4] cbeer opened pull request #186: Renamed FedoraDatastreams to generic FedoraBatch (master...http-batch-refactor) http://git.io/ZtrJXA
* github-ff leaves
<tecoripa>no, more like object not validating...
<awoods>nbanks/cbeer: hmm, I am not sure I see the point in the "mixin" query param on POST object, then.
<pivotal-bot>Chris Beer added comment: "Fixed by https://github.com/futures/fcrepo4/pull/186" https://www.pivotaltracker.com/story/show/60876990
<tecoripa>probably be a good idea to get cbeer to chime in16:42
<awoods>mikeAtUVa: Where does this ticket stand? https://www.pivotaltracker.com/story/show/61421690
<pivotal-bot>feature: Add alternate versioning policy. (finished) / owner: Mike Durbin
<nbanks>awoods: We can do it via import so I'm fine with that for my use case.
<pivotal-bot>Chris Beer edited "Fix up fcr:datastreams for a path-based world" https://www.pivotaltracker.com/story/show/60876990
<tecoripa>mikeAtUVa: yes, I think that's it exactly: create the complete object, THEN add the node type to it.16:43
<pivotal-bot>Chris Beer added comment: "After this is merged, I'll add the copy from the design doc and javadoc to the REST API." https://www.pivotaltracker.com/story/show/60876990
<cbeer>awoods: yes, it looks like FedoraNodes limits you to object or datastream (our core types)
i think we generally infer the right-thing-to-do by applying some heuristics to the content type
<nbanks>awoods: I think in this case My CND is also wrong as I was extending fedora:object rather than just defining other mixin types that could be used in conjunction with each other.16:44
<cbeer>but you might, e.g., want to add a datastream with an RDF payload.. and to do that, you need to help FedoraNodes out.
<tecoripa>mikeAtUVa, nbanks: I still think cmodel node types should be subtypes of fedora:resource. since they'll constrain both objects and datastreams
<cbeer>(or, of course, maybe even preferably, use the fcr:content endpoint)
<tecoripa>but I'm willing to be convinced otherwise
<cbeer>tecoripa: i think you should be able to add a mixin when creating a new node by posting some RDF or sparql-update queries as the payload16:45
<awoods>cbeer: But we have HTTP endpoints for creating objects and datastreams. What purpose does the mixin query param serve?
<cbeer>tecoripa: i'm not sure you could do both that and upload content though.
* tecoripa nods along with awoods
<cbeer>awoods: if i were to guess, it predated fcr:content in some way.
<nbanks>tecoripa: I'm not sure they need to extend fedora:resource, that would cause to mixin the same definition twice
<cbeer>awoods: i suggested earlier today replacing the explicit mixin parameter with some content-disposition http magic.16:46
<nbanks>cbeer: "but you might, e.g., want to add a datastream with an RDF payload.. and to do that, you need to help FedoraNodes out." is this a limitation of importing?
<cbeer>nbanks: we assume any RDF payload posted to FedoraNodes should be applied to the newly created node, not imported as binary content.16:47
<tecoripa>nbanks: I don't follow -- same definition twice?
<mikeAtUVa>nbanks: have you observed problems adding two mixins which extend the same?
<nbanks>tecoripa & mikeAtUVa: I'm sure it does no harm, but it's just redundant
cbeer: thanks for the clarification16:48
<cbeer>(awoods: and, to be clear, i'm not defending ?mixin.. i'm just telling you why we have it.)16:49
<awoods>cbeer: If we have a better way of adding mixins to objects, that is great (content-disposition). Currently, it is not clear to me how we add custom mixins to existing objects at all.
<pivotal-bot>Chris Beer finished "Fix up fcr:datastreams for a path-based world" https://www.pivotaltracker.com/story/show/60876990
<mikeAtUVa>nbanks: I see it as saying something is a "man" and a "human" when both of those extend from "animal" It allows them to be used separately... that said, since everything (other than content) in our repo *HAS* to be a "fedora:object" or "fedora:datastream", you're right that it'll be redundent.
<tecoripa>nbanks: ah, I see -- if you attach fedora:object and myns:myNodeType to an object, and they're both subtypes of fedora:resource, then you're syaing there's duplication...
<cbeer>awoods: isn't that what mikeAtUVa documented earlier this week for you?
<nbanks>tecoripa: yup
<tecoripa>nbanks, mikeAtUVa: then it seems that custom cmodel node types should be subtypes of fedora:object16:52
but then should datastream constraints, customarily also defined in a cmodel, be broken out into node types that are subtypes of fedora:datastream?16:53
<awoods>cbeer: Yes, in a way, maybe. It looks like mikeAtUVa's approach adds the mixin to all object resources, as opposed to a single object. mikeAtUVa?
<cbeer>awoods: :57 ] mikeAtUVa> esme,awoods,cbeer: the sparql for adding that mixin does work and will find its way into an integration test I'm making and the corresponding walkthrough for versioning that describes one way to create auto-versioning objects."
* travis-ci joins16:54
[travis-ci] futures/fcrepo4#1334 (http-batch-refactor - 9035558 : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/commit/90355589814b
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14947687
* travis-ci leaves
<mikeAtUVa>cbeer, awoods: yeah the other approach is in an integration test, but not yet in documentation. It's not clear to me that it's a useful case for versioning since I discovered the neat override works.16:55
<awoods>mikeAtUVa: By "neat override", you mean the sparql update?
<nbanks>tecoripa: I don't think we have to extend anything with our cmodels. The cmodel is just a further constraint on fedora:object and fedora:datastreams, it looks like a cmodel definition will be split out across several types though since you will have to define constraints on your child datastreams separately.
<mikeAtUVa>awoods: I mean posting a CND fragment that replaces the existing definitino.16:56
awoods: a a feature, what's valid, invalid and what happens when you post node type definitions that have the same URI as existing ones is not yet fully explored... to my knowledge.16:57
awoods: that you can cleanly replace one is handy because it works in this case, though I don't know what happens to all of the objects that may or may not be valid with the new definition.
<tecoripa>nbanks: errk, that's what I was afraid of. So A Fedora 3 CModel which defines three datastreams becomes for nodeTypes: one for the object, and three for the datstreams.16:59
<cbeer>mikeAtUVa: i'm 90% sure modeshape doesn't let you make changes that would make an object invalid.
<mikeAtUVa>cbeer: I would believe that... I tried lots of CND updates that gave me odd NPEs somewhere in the validation logic.. wasn't sure if it was malformed CND, illegal changes or what...17:00
<bljenkins>Project fcrepo-fixity-corrupter build #525: SUCCESS in 59 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/525/
<nbanks>tecoripa: Thats how I think it works, but you don't really need to describe the children if they don't do anything special like have nested datastreams.
tecoripa: sorry not children I meant datastreams.
cbeer: "I'm 90% sure modeshape doesn't let you make changes that would make an object invalid." This is what I read in the doc's it should only allow you to make non-destructive changes.17:01
<bljenkins>Project fcrepo-kitchen-sink build #695: STILL FAILING in 1 min 24 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/695/
<pivotal-bot>Andrew Woods added comment: "Where does this ticket stand, @md?" https://www.pivotaltracker.com/story/show/61421690
<tecoripa>nbanks: I'm wondering how we would implment the description case. I want to say, "This node is type FOO; it will have three datastreams, BAR1, BAR2, and BAR3; BAR1 and BAR2 are required, but BAR3 is optional."17:02
<cbeer>tecoripa: OWL?17:03
<tecoripa>Isee how you can say that a datastream is required in an object CND: +mynsBAR1(mynsBAR1Type)
but optional?
<nbanks>tecoripa: All datastreams are optional except the ones that say "mandatory" like so + ns:BAR (fedora:datastream) = nt:unstructured mandatory17:04
<tecoripa>cbeer: are we really going to ask users to write OWL ontologies for their object types?
<pivotal-bot>Mike Durbin added comment: "I'm waiting 10+ minutes for mvn clean install to verify that no more tests broke when I reworked the methods..." https://www.pivotaltracker.com/story/show/61421690
<cbeer>tecoripa: we made them write XACML and WSDL in fcrepo317:05
<tecoripa>nbanks: ah, OK
cbeer: not exactly a strong argument in your favor, I'd say
<mikeAtUVa>cbeer, tecoripa: we could always use earlier versions of fedora client to write the XACML and WSDL for us.
<nbanks>cbeer: let the frustration continue?17:06
<tecoripa>I say we volunteer cbeer: he'll write everyone's OWL.
<nbanks>gtg supper17:11
<tecoripa>nbanks: before you leave...\
<pivotal-bot>Mike Durbin added comment: "Done, tests passed PR updated..." https://www.pivotaltracker.com/story/show/6142169017:12
<tecoripa>so your problem became a non-problem by deciding that we shouldn't be extending node types, just creating standalone mixins, correct?
<cbeer>tecoripa: his problem is the redundancy?17:13
* mikeAtUVa leaves17:14
<tecoripa>cbeer: I don't think so, at least not initially. he was unable to create any other types that extended fedora:object
> nbanks: awoods: It works as document in that you can create objects/datastreams but it seems as though I can't create any other registered types that extend the fedora:object type.17:15
<awoods>tecoripa: on a call
<pivotal-bot>Chris Beer started "MODE-2098" https://www.pivotaltracker.com/story/show/6105106017:26
* ermadmix leaves
* ermadmix joins17:27
<tecoripa>cbeer: do you if in our CND files, we need to put the fedora namespace declarations?17:43
or is that inherited? although this is for declaring a standalone mixin type, not a subtype17:44
do you know?
<cbeer>tecoripa: as long as your CND is loaded after the fedora one, you don't need to redefine it
<tecoripa>is there any guarantee what order it will be loaded in? seems like it might be a good idea to put it in there, just in case17:45
wouldn't hurt, right?
<cbeer>don't know.
<tecoripa>I'll put it in, see what happens.
* ksclarke leaves17:49
<nbanks>tecoripa: Sorry, ya my original problem was with extending the fedora types, but now that I've seen the error's or my ways.17:52
check out my half-baked page: https://wiki.duraspace.org/display/FF/How+to+Migrate+a+Fedora+3+Content+Model+to+a+Fedora+4+Node+Type
I'm fleshing out a real test case right now.
<nbanks>tecoripa: For datastream BIB1, you could write that the dc:fields are covered already by fedora:resource properties17:55
brings up a question that occurs to me now:
in cmodels, you can specify that datastream format: dublin core XML, for example
how is that done in the CND world? Or does that even make sense?
all I'm seeing are namespaced definitions of properties17:58
by not constraints on the contents themselves.
<nbanks>tecoripa: There are some default types that define mime type as a property but I don't think you can enforce that the content be of that mime/type
<tecoripa>but no constarints
I think that's what I want: mime type as a property17:59
<awoods>tecoripa: I guess there are two questions there: how do you describe what a datastream should be, and how do you validate that it is that thing.
<nbanks>tecoripa: Well you can enforce that the property "mime-type" have a specific value but the content will not be checked.
<tecoripa>that's all we have with Fedora 3, so that's okay18:00
<nbanks>tecoripa: jcr:content that is.
<tecoripa>node types are simply about structure, not content
<nbanks>tecoripa: I'll find the types one sec.
tecoripa: Oh fedora:datastream defines a child node image:metadata that defines mime type as a property. Not exactly what your looking for though, not sure why that's a supported default.18:01
<tecoripa>where are you seeing that CND? is that coming from nt:unstructured?18:02
here's what I see for fedora:datastream -> + image:metadata (nt:unstructured) multiple18:03
<nbanks>tecoripa: https://github.com/futures/fcrepo4/blob/master/fcrepo-kernel/src/main/resources/fedora-node-types.cnd
<tecoripa>nbanks: yep, that's what I'm looking at...18:04
but I don't see any reference to mime-type properties there. just another node type (nt:unstructured)18:05
<nbanks>tecoripa: Sorry for the delay seems like my internet is dying on me,18:07
<pivotal-bot>Eric James added "fcr:transform exception" https://www.pivotaltracker.com/story/show/61942638
<tecoripa>nbanks: not sure if you saw my comment: " but I don't see any reference to mime-type properties there. just another node type (nt:unstructured)"18:08
<nbanks>tecoripa: Sorry your right, I just seen the name of that datastream used as a type somewhere else and I was conflating the two.
<tecoripa>nbanks: from what I'm reading, it seems that mime-type of content is not really supported natively.
the only thing I can think of is to add a string property, something like "fedora:mime-type"18:09
which will probably make ajs6f wince, with good reason.
<nbanks>tecoripa: I'm pretty sure I seen it in the docs for the build in types, the page just doesn't want to load though :(
tecoripa: There we go https://docs.jboss.org/author/display/MODE/Built-in+node+types
<tecoripa>ah, I see: there is a jcr:mimeType property18:10
and there's even a mix:mimeType node type18:11
ok, I need to think about how to implement this...
do we tell users that they need to attach a mix:mimeType node type to their datastream objects?18:12
and then define the jcr:mimeType and jcr:encoding propertiesw on their datastream objects?
that seems the cleanest way to go
<nbanks>tecoripa: Hmm I don't think you need to for content, damn I can't create a datastream through the html interface, seems to be broken.18:13
<tecoripa>no, but they should add the mix:mimeType to any object that has the fedora:datastream node type18:14
<nbanks>tecoripa: nah ya you do, datastreams are nt:file not nt:resource
<tecoripa>as distinct from jcr:content (or whatever it is that contains the content)
nbaks: are we talking about the same thing? I'm thinking there's an object with the node type fedora:object, which has a child object with the node type fedora:datastream; that fedora:datastream object has a property jcr:content, which contains the actual bitstream18:17
is that correct so far?
<nbanks>tecoripa: So you have something like this for a datastream under a object + fedora:DC (fedora:datastream, mix:mimeType) = nt:unstructured
tecoripa: Pretty close fedora:datastream has a primary type of nt:file by default which has a mandatory child of type jcr:content18:18
<tecoripa>right, okay...
I see where you're going with this...
<nbanks>tecoripa: Ya so basically in the objects definition you need to add those mixins as required types.18:19
{+} prefixedName ( requiredTypes ) = defaultType attributes
<cbeer>tecoripa: sorry, i'm jumping in the middle of this.. your statement that "they should add the mix:mimeType to any object that has the fedora:datastream node type" doesn't make sense to me. mimeType is a property of the content (nt:resource), not the containing nt:file
<tecoripa>exactly. so then, I don't need to define the mime type at all in the datastream node type.18:20
<cbeer>or you'll get to the same place.
<nbanks>cbeer: But datastreams are of type nt:file, not nt:resource?
<cbeer>nbanks: right. the binary content is distinct from the fedora datastream.
(and, even then, we're collapsing a layer that jcr provides us...)18:21
<nbanks>cbeer: How does nt:resource fit in if the only file of nt:file is jcr:content? Does jcr:content extend jcr:resource? I think I'm lost.18:22
* ermadmix leaves
<nbanks>only file should be only child
<cbeer>a fedora:datastream is an nt:file
nt:file says there's a node at jcr:content that is a fedora:resource
sorry, that says it is a nt:resource18:23
and an nt:resource contains a binary property
at jcr:data (i think)
<tecoripa>cbeer: I'm making this up as I go along, so any insight you have would be great. basically, I'm trying to come up with a way to express in CND that the content of the datastream (whatever node or property that is) will have a mime type of X/X. Again, this is a descriptive assertion, not a normative one. just like in Fedora 3.
* github-ff joins18:24
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/Sv6W2A
fcrepo4/master d34748a Michael Durbin: Implemented auto versioning policy, made explicit versioning the default...
* github-ff leaves
<cbeer>tecoripa: and i'll predictably say that CNDs probably aren't where you want to have descriptive assertions.
<pivotal-bot>Andrew Woods added comment: "Resolved with: https://github.com/futures/fcrepo4/commit/d34748acfad255250ac115f3e7a4190de8dfb276" https://www.pivotaltracker.com/story/show/6142169018:25
Andrew Woods delivered "Add alternate versioning policy." https://www.pivotaltracker.com/story/show/61421690
<tecoripa>cbeer: hmmm, okay, where should they go then? or should we dispense with them entirely, at least in our node type definitions?
<nbanks>Where is it defined that jcr:content is nt:resource? I don't see it from the ModeShape CND in the wiki?
* github-ff joins
[fcrepo4] awoods closed pull request #174: Implemented auto versioning policy, made explicit versioning the default... (master...versioning-6) http://git.io/Sp0ICw
* github-ff leaves
<cbeer>nbanks: https://github.com/ModeShape/modeshape/blob/master/modeshape-jcr/src/main/resources/org/modeshape/jcr/jsr_283_builtins.cnd
i guess it's only an nt:resource by convention?
<nbanks>cbeer: I don't see it explicitly here, just that the default type is nt:base which could be anything
<tecoripa>cbeer: or to put it another way: what kinds of nodes would the mix:mimeType mixin be applied?18:27
<nbanks>cbeer: Ok, so as long as it's guaranteed
tecoripa: the jcr:content node
tecoripa: By convention jcr:content is of type nt:resource which has the mix:mimeType mixin, there for it also has the properties you seek.18:28
cbeer: right?18:29
<tecoripa>nbanks: ok. and is the jcr:content node a child of a fedora:datstream node: or is fedora:datastream just another mixin type alongside jcr:content?
<nbanks>tecoripa: jcr:content is a child node of fedora:datastream
<cbeer>we're guaranteeing that for fcrepo4 at least.
<nbanks>cbeer: Cool, maybe I should document that somewhere or maybe it already is and I missed it.18:30
<tecoripa>ok, so our datastream type node COULD declare that it has a childnode of the types jcr:content and mix:mimeType18:33
<nbanks>tecoripa: your datastream already has a child node jcr:content which has nt:resource as a primary type by convention. nt:resource already has mix:mimeType as a mixin so you don't need to define it.18:35
<tecoripa>OK. I'll put thsat in the wiki page. By virtue of assigning a node type fedora:datastream to a node, they would get the jcr:mimeType property for free.18:36
<nbanks>tecoripa: Awesome
<tecoripa>now, moving from descriptive to normative, which will make cbeer happier...18:37
can we say in our cnd that the jcr:mimeType property is mandatory, and must have the value "x/x"?
<nbanks>nbanks: My guess would be yes, that the further refinements would be enforced but, I think it warrant's testing.18:38
<tecoripa>right. and per our discussion earlier, you'd want to apply this nodeType to the object only after creating all the child fdatastream nodes, with the appropriate properties, etc.18:39
* travis-ci joins
[travis-ci] futures/fcrepo4#1337 (master - d34748a : Michael Durbin): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/944748801458...d34748acfad2
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14953183
* travis-ci leaves
<nbanks>tecoripa: I'm not sure, I haven't tested what would happen in such a case. If Mike was around I'd ask him to chime in.18:41
tecoripa: I think he said that for all mandatory properties and children must be in place before changing the type.18:42
<tecoripa>nbanks: I think cbeer confirmed that last week, too
nbanks: in any event, I just want to try and get a usable CND defined right now. if you're still around in a few minutes, I'll have you take a look at the CND for the BIB0DatastreamType.18:43
<nbanks>tecoripa: sure18:44
<bljenkins>Project fcrepo-fixity-corrupter build #526: SUCCESS in 1 min 35 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/526/18:46
<tecoripa>nbanks: looks like you have to specify the child node NAME in the CND...
which could be anything, right?
I'd prefer just to say: "a child with any node name, but with the required types jcr:content and mix:MimeType"18:47
<nbanks>tecoripa: yup, or provide a wild card which means it can have a child with any name of X type.
<tecoripa>got it: "The node name must be a valid JCR name or *, to indicate a residual child node definition. It may be single-quoted."18:48
<bljenkins>Project fcrepo-kitchen-sink build #696: STILL FAILING in 1 min 58 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/696/
<tecoripa>ok, now I'm stuck at setting a mandatory jcr:mimeType property on the jcr:content node18:58
or maybe I shouldn't bother with defining that...18:59
the only solution I see for that is to create a THIRD node type, to mixin with jcr:content,19:01
and define the mandatory property value there.
so to implement something resembling a CModel, you'd need node types at three levels: object, datastream, and content
mixins alongside fedora:object, fedora:datastream, and jcr:content19:02
<nbanks>tecoripa: Ya we need types at all three levels. So you define your custom node type and its custom datastream say fedora:DC.19:10
<tecoripa>nbanks: so then we'd have a content node type that would have the mandatory property jcr:mimeType, with the mandatory value "text/xml".19:12
<nbanks>tecoripa: makes sense to me19:13
<tecoripa>OK, this is coming clearer. and our datastream would require a node with that type.
our datastream node type, I mean.19:14
offhand, do you know of any way to also set a format-uri property?19:15
<nbanks>I think this might work?
[fedora:DCContent] mix
- jcr:mimeType (string) < 'text\/xml'
+ jcr:content (fedora:DCContent) = nt:unstructured
<tecoripa>do we need to define a default type?19:16
<nbanks>tecoripa: It depends on if Modeshape will combine the mixin's
<tecoripa>jcr:content is the type, not the name, right?
<nbanks>tecoripa: jcr:content is the name19:17
<nbanks>tecoripa: What do you mean by format-uri property? Like you want the regex for it?19:18
<tecoripa>here';s what we have in dsTypeModel: <form FORMAT_URI="http://www.openarchives.org/OAI/2.0/oai_dc" MIME="text/xml">
<pivotal-bot>Andrew Woods accepted "Add alternate versioning policy." https://www.pivotaltracker.com/story/show/6142169019:20
<nbanks>[fedora:DCContent] mix
- jcr:mimeType (string) < 'text/xml'
- fedora:formatURI (URI) < 'http://www.openarchives.org/OAI/2.0/oai_dc'
+ jcr:content (fedora:DCContent) = nt:unstructured
<tecoripa>ah, clever...
where did you get fedora:formatURI from?19:22
not seeing that in the fedora CND
* nbanks leaves19:34
* nbanks joins
* nbanks leaves19:35
* nbanks joins19:36
I just made it up :P
I gtg, see ya tomorrow
* nbanks leaves
<tecoripa>nbanks: yeah I figured that out. :)
<pivotal-bot>Andrew Woods accepted "jms-indexer-pluggable - comply to header messaging" https://www.pivotaltracker.com/story/show/61750308
* tecoripa leaves20:05
<bljenkins>Project fcrepo-jms-indexer-pluggable build #315: ABORTED in 1 day 3 hr: http://ci.fcrepo.org/jenkins/job/fcrepo-jms-indexer-pluggable/315/20:14
escowles: Non-functional update that makes tests less brittle, introducing a new TestIndexer class that allows ITs to check whether messages have been received without having to count files/triples.
Project fcrepo-jms-indexer-pluggable build #316: STILL UNSTABLE in 7 min 32 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-jms-indexer-pluggable/316/20:22
A. Soroka: refactor IndexerGroup for new messaging format, adding timeouts to tests, stronger Sparql indexer
* ksclarke joins20:27
<pivotal-bot>Andrew Woods added "fcrepo-jms-indexer-pluggable Build Errors" https://www.pivotaltracker.com/story/show/6195036221:09
Andrew Woods started "fcrepo-jms-indexer-pluggable Build Errors" https://www.pivotaltracker.com/story/show/61950362
* github-ff joins21:14
[fcrepo-jms-indexer-pluggable] awoods pushed 1 new commit to master: http://git.io/t_053w
fcrepo-jms-indexer-pluggable/master 5e7c81f Andrew Woods: Fix build errors...
* github-ff leaves
<pivotal-bot>Andrew Woods added comment: "- Enable project to dynamically define jms.port
- Remove erroneous bean property in indexer-core.xml
Resol..." https://www.pivotaltracker.com/story/show/61950362
Andrew Woods delivered "fcrepo-jms-indexer-pluggable Build Errors" https://www.pivotaltracker.com/story/show/61950362
<bljenkins>Yippie, build fixed!21:23
Project fcrepo-jms-indexer-pluggable build #317: FIXED in 9 min 32 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-jms-indexer-pluggable/317/
awoods: Fix build errors
<pivotal-bot>Andrew Woods accepted "fcrepo-jms-indexer-pluggable Build Errors" https://www.pivotaltracker.com/story/show/6195036221:24
* travis-ci joins
[travis-ci] futures/fcrepo-jms-indexer-pluggable#41 (master - 5e7c81f : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo-jms-indexer-pluggable/compare/4a41346d8e98...5e7c81fd0a55
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo-jms-indexer-pluggable/builds/14959748
* travis-ci leaves
* tecoripa joins21:41
* tecoripa leaves
* tecoripa joins
* tecoripa leaves21:42
<pivotal-bot>Andrew Woods added comment: "I saw a failure with the FedoraHtmlResponsesIT, but subsequent builds passed. We may want to keep an eye on..." https://www.pivotaltracker.com/story/show/6191498021:53
Andrew Woods delivered "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/6191498021:54
* github-ff joins
[fcrepo4] awoods pushed 2 new commits to master: http://git.io/yqX87g
fcrepo4/master d3a495c Nigel Banks: Add Namespace in browser now functioning, also added an integration test
fcrepo4/master fa517f2 Andrew Woods: Merge pull request #185 from nigelgbanks/61914980-namespace-prefix...
* github-ff leaves
* travis-ci joins22:07
[travis-ci] futures/fcrepo4#1338 (master - fa517f2 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/d34748acfad2...fa517f255803
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/14961132
* travis-ci leaves
<bljenkins>Project fcrepo4 build #1491: UNSTABLE in 1 hr 13 min: http://ci.fcrepo.org/jenkins/job/fcrepo4/1491/23:07
nigel.g.banks: Add Namespace in browser now functioning, also added an integration test
<pivotal-bot>Andrew Woods edited "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/6191498023:37
Andrew Woods accepted "Creating Namespace via the HTML interface doesn't work." https://www.pivotaltracker.com/story/show/61914980

Generated by Sualtam