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

Using timezone: Eastern Standard Time
* ksclarke leaves00:59
* dwilcox joins01:35
* dwilcox leaves01:49
* dwilcox joins03:08
* dwilcox leaves03:34
* dwilcox joins07:26
* fasseg joins07:37
* dwilcox leaves07:56
* dwilcox joins07:58
* dwilcox leaves08:12
* dwilcox joins08:17
* mikeAtUVa joins08:28
* mohamed joins08:32
* awead joins08:47
* ksclarke joins09:03
* dwilcox leaves09:04
<pivotal-bot____>Andrew Woods edited "Improve object/byte count performance, or move to separate REST endpoint" https://www.pivotaltracker.com/story/show/7864724809:31
* fasseg leaves09:37
* gregjansen joins
* gregjansen leaves09:38
<pivotal-bot____>A. "Banjaxel" Soroka added "Too many compiler warnings" https://www.pivotaltracker.com/story/show/7927644609:45
A. "Banjaxel" Soroka started "Too many compiler warnings" https://www.pivotaltracker.com/story/show/79276446
* github-ff joins
[fcrepo4] ajs6f opened pull request #482: Cleaning up Java compiler warnings (master...TooManyExceptions) http://git.io/e-rbjA
* github-ff leaves
<pivotal-bot____>A. "Banjaxel" Soroka added comment: "https://github.com/fcrepo4/fcrepo4/pull/482" https://www.pivotaltracker.com/story/show/79276446
A. "Banjaxel" Soroka finished "Too many compiler warnings" https://www.pivotaltracker.com/story/show/79276446
Andrew Woods added "Fixity result status missing from admin console" https://www.pivotaltracker.com/story/show/7927698009:51
* awead leaves09:52
* awead joins09:54
* gregjansen joins09:55
* gregjansen leaves09:56
* mohamed1 joins09:58
<pivotal-bot____>Andrew Woods delivered "Too many compiler warnings" https://www.pivotaltracker.com/story/show/7927644610:00
* github-ff joins
[fcrepo4] awoods pushed 2 new commits to master: http://git.io/rUhBpA
fcrepo4/master 9ccde9c ajs6f: Cleaning up Java compiler warnings
fcrepo4/master 6197922 Andrew Woods: Merge pull request #482 from ajs6f/TooManyExceptions...
* github-ff leaves
* mohamed leaves10:01
* ajs6f joins
awoods; Do you want to do an intermediate merge for that i-test id thing?10:07
<awoods>ajs6f: You can continue to merge with master, no? until it is complete?10:10
<ajs6f>awoods: Yeah, I just know this is going to take a really long time (lots of churn). It's fine either eay.
* travis-ci joins10:11
[travis-ci] fcrepo4/fcrepo4#2395 (master - 6197922 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/fcrepo4/fcrepo4/compare/f374560921d3...6197922d6bc2
[travis-ci] Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/35938759
* travis-ci leaves
<awoods>ajs6f: If you want to split it into separate tickets, I can push them in turn. If you want to work it as one big ticket, I will wait.
<ajs6f>awoods: Okay, I'll split it later today. I want to look at the weirdness with RdfStream that @cbeer turned up. It smells like it might be something deeper and RdfStream is a very low-level type, so I want it to be completely reliable.10:12
<awoods>ajs6f: thanks10:13
<ajs6f>awoods: No, thank you. For all that you do every day in so many little ways.
* scossu joins10:22
* dwilcox joins10:25
* mohamed1 leaves10:44
* dwilcox leaves10:47
* ajs6f1 joins10:51
* ajs6f leaves
* ajs6f1 leaves10:52
* whikloj joins10:54
* ajs6f joins11:00
* mohamed joins11:02
<pivotal-bot____>A. "Banjaxel" Soroka added "Get #fcrepo IRC channel back." https://www.pivotaltracker.com/story/show/7927841211:06
Andrew Woods added "CSS warnings when build fcrepo-webapp" https://www.pivotaltracker.com/story/show/79278426
A. "Banjaxel" Soroka edited "Get #fcrepo IRC channel back." https://www.pivotaltracker.com/story/show/79278412
Andrew Woods edited "Velocity templates should be called for ascenedant primary types" https://www.pivotaltracker.com/story/show/79207804
Andrew Woods accepted "Too many compiler warnings" https://www.pivotaltracker.com/story/show/7927644611:10
Andrew Woods accepted "Final walk-through of Metrics user docs" https://www.pivotaltracker.com/story/show/78690924
Andrew Woods accepted "Refactor FedoraResource#get*Triples into a single getTriples method" https://www.pivotaltracker.com/story/show/78960386
Andrew Woods accepted "Ensure a clean shutdown of the repository on container shutdown." https://www.pivotaltracker.com/story/show/66454096
Andrew Woods accepted "If someone tries to create a new node with an unregistered namespace in the HTML UI, capture that error and give th�" https://www.pivotaltracker.com/story/show/61500426
Andrew Woods added "Enable fcrepo4-release-tests to build from clean .m2" https://www.pivotaltracker.com/story/show/7928649411:29
Andrew Woods started "Enable fcrepo4-release-tests to build from clean .m2" https://www.pivotaltracker.com/story/show/79286494
<cbeer>ajs6f: is there some reason we don't use class-level @Contexts, e.g. https://gist.github.com/cbeer/550c13988360f00d8ed1?
<ajs6f>cbeer: I know of none.11:30
In fact, I remember wondering about that myself.
<cbeer>ajs6f: ok. i'll do it for FedoraNodes and FedoraContent at least11:31
<ajs6f>Cool. That should tighten the code nicely.
<pivotal-bot____>Andrew Woods added comment: "https://github.com/fcrepo4-labs/fcrepo4-release-tests/pull/1" https://www.pivotaltracker.com/story/show/7928649411:35
Andrew Woods finished "Enable fcrepo4-release-tests to build from clean .m2" https://www.pivotaltracker.com/story/show/79286494
Andrew Woods added "SPARQL Recipes failing on beta-4-snapshot" https://www.pivotaltracker.com/story/show/7928721811:37
Andrew Woods added comment: "Moved to fcrepo4-labs: ""
https://github.com/fcrepo4-labs/fcrepo4-release-tests" https://www.pivotaltracker.com/story/show/72524140
Andrew Woods delivered "Automate running sqarql-recipes" https://www.pivotaltracker.com/story/show/72524140
<cbeer>java merge conflicts--11:38
<ajs6f>Haskell merge conflicts ++
That doesn't work either.
* gregjansen joins11:39
<awoods>cbeer: can your rebase PR-481 on master?11:41
<cbeer>awoods: working on it now. hence the complaint about java merge conflicts.11:42
<awoods>cbeer: perfect... I guess
<pivotal-bot____>Andrew Woods deleted "SPARQL Recipes failing on beta-4-snapshot" https://www.pivotaltracker.com/story/show/7928721811:43
* gregjansen leaves11:44
<cbeer>awoods: i guess. too bad ajs6f spent time cleaning up methods i've killed, though.11:45
That's one good reason for me to stop fooling with the i-tests until you are done.
<f4jenkins>Yippee, build fixed!11:47
Project fcrepo4 build #2158: FIXED in 14 min: http://jenkins.fcrepo.org/job/fcrepo4/2158/
<cbeer>ajs6f: hopefully after 481, i'll be done with messing with -kernel for a bit.11:48
<ajs6f>cbeer: It's worth it. You're doing good stuff.11:49
* github-ff joins
[fcrepo4] cbeer force-pushed new-interfaces from d1fd14e to adcb75d: http://git.io/77CTdQ
fcrepo4/new-interfaces 7f1241e Chris Beer: Extract a FedoraBinary interface/impl out of FedoraDatastream
fcrepo4/new-interfaces 2174c91 Chris Beer: Simplify DatastreamService and ObjectService contracts
fcrepo4/new-interfaces 3c27fc8 Chris Beer: Remove NodeService#findOrCreateObject
* github-ff leaves
* github-ff joins11:50
[fcrepo4] cbeer force-pushed new-interfaces from adcb75d to a8dd0b4: http://git.io/77CTdQ
fcrepo4/new-interfaces a8dd0b4 Chris Beer: Pull FedoraContent @Context parameters to the class level
* github-ff leaves
<cbeer>ajs6f: it's nice to have a cohesive set of core modules. there's a ton of work that should go into the *Service still.11:51
but i don't understand versioning or locking well enough to take a hatchet to it.
<f4jenkins>Yippee, build fixed!
Project fcrepo-module-auth-rbacl build #127: FIXED in 4 min 13 sec: http://jenkins.fcrepo.org/job/fcrepo-module-auth-rbacl/127/
<ajs6f>cbeer: Yep. We're going to have an easier time providing a coherent and well-modularized HTTP API if we have a coherent and well-modularized Java API.
* travis-ci joins11:52
[travis-ci] fcrepo4/fcrepo4#2396 (new-interfaces - adcb75d : Chris Beer): The build has errored.
[travis-ci] Change view : https://github.com/fcrepo4/fcrepo4/compare/d1fd14e408f1...adcb75d0a811
[travis-ci] Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/35950171
* travis-ci leaves
<cbeer>awoods: rebased.
<f4jenkins>Project fcrepo-message-consumer build #672: UNSTABLE in 5 min 22 sec: http://jenkins.fcrepo.org/job/fcrepo-message-consumer/672/11:53
<awoods>cbeer: thanks, I will give it a look
<pivotal-bot____>Andrew Woods accepted "Acceptance test: Admin console" https://www.pivotaltracker.com/story/show/7882295411:54
<f4jenkins>Project fcrepo4 build #2159: UNSTABLE in 11 min: http://jenkins.fcrepo.org/job/fcrepo4/2159/12:00
* nikhiltri joins12:03
* travis-ci joins12:05
[travis-ci] fcrepo4/fcrepo4#2397 (new-interfaces - a8dd0b4 : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/fcrepo4/fcrepo4/compare/adcb75d0a811...a8dd0b4b1d28
[travis-ci] Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/35950266
* travis-ci leaves
<f4jenkins>Yippee, build fixed!12:16
Project fcrepo4 build #2160: FIXED in 15 min: http://jenkins.fcrepo.org/job/fcrepo4/2160/
Yippee, build fixed!12:23
Project fcrepo-message-consumer build #673: FIXED in 6 min 38 sec: http://jenkins.fcrepo.org/job/fcrepo-message-consumer/673/
* mohamed leaves12:27
* github-ff joins12:30
[fcrepo4] ajs6f opened pull request #483: Factor out LDP prefer mechanics, Guava-fy FedoraNodes.describe() (master...RdfStreamWeirdness) http://git.io/iaOqbg
* github-ff leaves
<ajs6f>cbeer: ^^^
That's for you to look at.
<cbeer>ajs6f: looks good to me.12:31
<ajs6f>Make a ticket if you want, and merge it?12:32
<cbeer>ajs6f: i'll drag it into my http api refactor
<cbeer>grr. i guess i can't do this without jersey 2 :/12:33
<ajs6f>Do what? The @Context stuff?
<cbeer>@Context was fine.. now i'm trying to do:12:34
public FedoraNodes(@PathParam("path") final List<PathSegment> pathList)
<ajs6f>I thought we already did that?
<cbeer>although i see it in the jersey 1.x docs too so.. maybe there's something funky with spring + jersey 1.x
<ajs6f>Oh, you're trying to do it in the constructor...
<cbeer>ajs6f: not the in the constructor.
<cbeer>ajs6f: ah, it was a scoping thing.. prototype vs request12:38
ajs6f: i assume we like prototype?
<ajs6f>cbeer: Depends on how much state we're carrying in those guys.12:39
If it's very little, the overhead is then just the cost of instantiating an object, which should be microscopic.
I would try @Request. I can't think of any real state in the JAX-RS controllers.
(And there really oughtn't be, anyway.)12:40
Sorry, the overhead above means "the overhead of dong request-scoping")
<cbeer>oh, no, i was wrong anyway. just got a totally different error.12:41
<cbeer>oh well. it was an idea.12:44
<ajs6f>That's how I feel about Fedora.12:45
* mohamed joins12:51
* github-ff joins13:11
[fcrepo4] nikhiltri opened pull request #484: Velocity templates should be called for ascenedant primary types (master...test-velocity-templates) http://git.io/mpeYtQ
* github-ff leaves
* fcrepo-bot joins13:14
* ajs6f leaves13:16
<pivotal-bot____>Nikhil Trivedi added comment: "I pushed up a pull request for this issue: https://github.com/fcrepo4/fcrepo4/commit/23d0a2db6379b7f3d78057�" https://www.pivotaltracker.com/story/show/7920780413:37
Nikhil Trivedi finished "Velocity templates should be called for ascenedant primary types" https://www.pivotaltracker.com/story/show/79207804
* nikhiltri leaves14:08
* nikhiltri joins14:09
<awoods>cbeer: with the exception of the two nodeTypes() methods, it is unclear why the other NodeService methods do not migrate to ObjectService.
<cbeer>awoods: it's true.
but also not on my critical path.
<awoods>cbeer: That makes sense. There is value in minimizing those such changes in your current refactoring.14:10
* github-ff joins14:15
[fcrepo4] awoods deleted new-interfaces at a8dd0b4: http://git.io/GYUXYw
* github-ff leaves
<pivotal-bot____>Andrew Woods delivered "Update core interfaces" https://www.pivotaltracker.com/story/show/79217018
<cbeer>the fact that i'm looking at jersey2 again proves that to be too true.14:17
* fcrepo-bot leaves14:21
<awoods>cbeer: ?14:22
* travis-ci joins14:31
[travis-ci] fcrepo4/fcrepo4#2401 (master - c7dc9f3 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/fcrepo4/fcrepo4/compare/6197922d6bc2...c7dc9f3a7b33
[travis-ci] Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/35965289
* travis-ci leaves
<f4jenkins>Project fcrepo-module-auth-xacml build #201: FAILURE in 34 sec: http://jenkins.fcrepo.org/job/fcrepo-module-auth-xacml/201/14:35
* scossu leaves14:45
* scossu joins14:51
<whikloj>mikeAtUVa: ping15:16
<mikeAtUVa>whikloj: hi15:17
<whikloj>mikeAtUVa: I'm having a problem, for some reason when I try to get the lock for a session and path, the token is always null. Is that normal?15:18
wikloj: My recollection is that the token was always something other than null.
whikloj: I haven't tested it in months though... I'll do so now...15:19
<whikloj>mikeAtUVa: Could be I am using it incorrectly.15:20
mikeAtUVa: I am using a LockServiceImpl.getLock() from the FedoraResourceImpl. It works as expected between lock or no lock. But the JCRLock object returned by lock.getLockToken() is null.15:21
mikeAtUVa: that should read I'm call LockService.getLock() from inside of FedoraResourceImpl
<mikeAtUVa>whikloj: If I recall correctly, those methods don't return you the lock token if it isn't held by your session...15:22
<whikloj>mikeAtUVa: Ok, that would make more sense.15:23
<mikeAtUVa>whikloj: maybe when we said we should inlcude it in the etag, we were too optimistic and we can really only include whether it's locked in the etag..15:24
<pivotal-bot____>Andrew Woods added comment: "In addition to code review comments, please add an integration test to: ""
https://github.com/fcrepo4/fcrepo4/bl�" https://www.pivotaltracker.com/story/show/79207804
Andrew Woods rejected "Velocity templates should be called for ascenedant primary types" https://www.pivotaltracker.com/story/show/79207804
<whikloj>mikeAtUVa: yeah, I'll stick something in just to alter the eTag value. thanks.15:25
awoods: about the caching precondition checking? It doesn't seem to care if If-None-Match tags don't match only if they do match. So I am getting a 304 from the server even though my If-None-Match etag is different.16:21
awoods: I am not confident enough to start tearing it apart though.16:22
<awoods>whikloj: Are you asking a question?16:25
whikloj: what call are you making?16:26
<whikloj>awoods: No more of a statement. Based on my understand of the spec, when a If-None-Match etag exists and is different the server should return the page. Jersey is not, it's sending a 304.16:27
awoods: So I'm not sure how to affect that or if I am way off the mark.
<awoods>whikloj: whichever method you are requesting must be making the 304 decision based on something besides If-None-Match
whikloj: have you looked at the source code?16:28
for whichever method?
<whikloj>awoods: yes, the code is the Jersey ContainerRequest checks the If-None-Match header, but only sets a 304. It seems to leave the rest up to us.16:30
awoods: https://github.com/fcrepo4/fcrepo4/blob/master/fcrepo-http-commons/src/main/java/org/fcrepo/http/commons/AbstractResource.java#L282
awoods: It returns builder as null, so we don't revalidate.
<awoods>whikloj: That seems odd, if the ContainerRequest detects a 304 situation, the builder it returns will not be null.16:35
<whikloj>awoods: Right, but it doesn't because the etag is different it doesn't set a 304. However we seem to only force a revalidate if we get a If-Match header and the etag is different.16:36
<awoods>whikloj: I am struggling to figure out what feedback you are looking for.16:39
<whikloj>awoods: why do we have the line I sent you. Why don't we always re-send the cached page unless Jersey sets the header to 304?16:40
awoods: sorry not "re-send" rather "rebuild"
<awoods>whikloj: The line you sent is only true if the builder is not null. The builder is only not null if there is a 304... so I am missing something in your question.16:44
cbeer: Do you have any thoughts here ^^16:45
<cbeer>awoods: summary?
<awoods>whikloj: summary?16:46
<whikloj>awoods/cbeer: If-None-Match etag sent by browser, does not match etag for item. However Jersey still seems to send a 304 along with the new eTag. Not sure why.16:47
<cbeer>whikloj: for what kind of request?
<whikloj>cbeer: GET
cbeer: actually, maybe I do. The etag reponse builder sends a null, then we check last modified date. It sets the 304. So even though we should rebuild the page based on the etag we cache based on date.16:50
awoods/cbeer: So long story (sorry) short, I can change the eTag based on a lock but it will not affect the page display in the HTML UI because the last modified date is not affected by the lock.16:52
* dwilcox joins16:55
<awoods>whikloj: and the same time that you are updating the etag (based on the presence of a lock) you are unable to update the last-modified-date?16:57
<whikloj>awoods: I am changing the eTag value when you check it. So I could change the last modified date but it would get updated everytime you check it. We could maybe set a new property for lockTime and use that in the check, but it would require a change to the locking code.17:01
awoods: As this is all for the display of locks on the HTML UI. So I didn't know how much of a change we wanted to implement.17:03
<awoods>whikloj: It is really for any client that relies on caching
<whikloj>awoods: Well then I guess the question is should we just have mikeAtUVa change the locking code to affect the lastModifiedDate of all objects that get locked? This would render the etag problem irrelevant/17:05
<awoods>whikloj: How would that happen? recursively add a lock-date to each child?17:06
* awead leaves17:07
<whikloj>awoods: Either actually update the last modified date, so now consider an item that is locked to have been altered. Or add a new property (recursively on deep locks), and alter the check for lastModifiedDate.17:08
<awoods>whikloj: I do not think we want to change the real last-modified dates...17:09
whikloj: I wonder if there is a date on the lock-object that we can use? http://www.day.com/maven/jcr/2.0/17_Locking.html#17.11.4 Testing for Lock Holding17:10
<whikloj>awoods: I'll have a look.17:11
<mikeAtUVa>whikloj: "alter the check for lastModifiedDate" means that that we'd return not when the object was modified, but instead the creation date for a held lock if one was held?
<awoods>mikeAtUVa: that is right, and as we have previously discussed, I do not believe we want to do that.
<whikloj>mikeAtUVA: theoretically, yes. When the lock was added and again when it is removed.17:12
<awoods>mikeAtUVa/whikloj: potentially a property "lock-datetime" could be added to the locking node, but that leaves ourselves open to book-keeping errors in the code.17:13
* mikeAtUVa leans towards accepting locks that are not properly represented in responses when cached.17:14
<awoods>mikeAtUVa: you are suggesting: punt?
<whikloj>awoods: It sounded like this was an edge case, so I was wondering like mikeAtUVa. How much effort is too much?17:15
<mikeAtUVa>I know it's unsatisfying whikloj, awoods, but this really only affects whether a user or client can tell whether a resource is currently locked... something that might change between when they look at a resource and act on it anyway.
so yeah, I guess I vote to punt... but I'm not sure I'm 100% clear on the details/implications of this lock-datetime approach.17:16
<awoods>mikeAtUVa/whikloj: at the very least, we have a better understanding of the issue. If users in the wild complain about the behavior, we can revisit it.17:17
<whikloj>awoods/mikeAtUVa: Ok, I'll write a blurb in the ticket about why its happening and why we are not fixing it right now.
and move on17:18
<awoods>whikloj: ok, mikeAtUVa: I could imagine a property on the lock-holding-node be added/removed with creation/removal of locks. http://www.day.com/specs/javax.jcr/javadocs/jcr-2.0/javax/jcr/lock/Lock.html#getNode()
mikeAtUVa: that property (lock-datetime) if present, could be returned as the lastmodified date in responses on locked nodes17:19
* github-ff joins17:22
[fcrepo-module-auth-xacml] awoods pushed 1 new commit to master: http://git.io/EVlykQ
fcrepo-module-auth-xacml/master 9922d93 Andrew Woods: Address impacts from fcrepo DatastreamService API refactoring...
* github-ff leaves
<pivotal-bot____>Andrew Woods added comment: "Related commit addressing impacts in fcrepo-module-auth-xacml: ""
https://github.com/fcrepo4/fcrepo-module-auth-�" https://www.pivotaltracker.com/story/show/79217018
<f4jenkins>Yippee, build fixed!17:24
Project fcrepo-module-auth-xacml build #202: FIXED in 2 min 19 sec: http://jenkins.fcrepo.org/job/fcrepo-module-auth-xacml/202/
awoods: Address impacts from fcrepo DatastreamService API refactoring
* mikeAtUVa leaves
<f4jenkins>Project fcrepo-message-consumer build #674: ABORTED in 2 hr 58 min: http://jenkins.fcrepo.org/job/fcrepo-message-consumer/674/17:28
* mohamed leaves17:29
<pivotal-bot____>Jared Whiklo added comment: "ETag value can be altered based on a lock, however this does not force a Responses.preconditi�" https://www.pivotaltracker.com/story/show/7704186817:32
Jared Whiklo finished "HTML UI sometimes displays out-of-date results." https://www.pivotaltracker.com/story/show/7704186817:33
Jared Whiklo started "Acceptance test: Backup/Restore" https://www.pivotaltracker.com/story/show/78826504
* dwilcox leaves17:45
* whikloj leaves17:59
<f4jenkins>Project fcrepo-message-consumer build #675: SUCCESS in 5 min 35 sec: http://jenkins.fcrepo.org/job/fcrepo-message-consumer/675/18:00
<pivotal-bot____>Andrew Woods added comment: "Could you please include the diff of your FedoraResourcdesImpl.java, so we know what was changed?" https://www.pivotaltracker.com/story/show/77041868
* scossu leaves
* nikhiltri leaves18:29
* ksclarke leaves18:39
* awead joins19:53
* awead leaves20:21
* awead joins20:34
<cbeer>barmintor_: ping? (although it's nothing that can't wait until tomorrow.)20:40
* awead leaves20:58
<cbeer>nm. RepositoryRuntimeException needs to subclass WebApplicationException to invoke sensible jersey2/hk2 error handling.21:05
* ksclarke joins21:11

Generated by Sualtam