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

Using timezone: Eastern Standard Time
* dwilcox leaves02:04
* awoods leaves03:39
* awoods joins06:20
* acoburn joins07:50
* dhlamb joins08:13
* mikeAtUVa joins08:15
* escowles joins08:16
* whikloj joins08:46
* osmandin joins08:49
* awead joins08:53
* jgpawletko joins08:59
* ksclarke joins09:00
* github-ff joins09:02
[fcrepo4] ajs6f pushed 1 new commit to BNodesAreBNodes: http://git.io/vvl6C
fcrepo4/BNodesAreBNodes 9afaa0d A. Soroka: Elaborating code for awoods
* github-ff leaves
* ajs6f joins09:18
* travis-ci joins
fcrepo4/fcrepo4#3560 (BNodesAreBNodes - 9afaa0d : A. Soroka): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/44d2343b1012...9afaa0d532be
Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/58436817
* travis-ci leaves
* dwilcox joins09:21
* github-ff joins09:23
[fcrepo4] ajs6f pushed 1 new commit to BNodesAreBNodes: http://git.io/vvlSP
fcrepo4/BNodesAreBNodes b1738b9 A. Soroka: Making code more verbose for awoods
* github-ff leaves
<f4jenkins>Project fcrepo-message-consumer build #1112: UNSTABLE in 5 min 23 sec: http://jenkins.fcrepo.org/job/fcrepo-message-consumer/1112/
* travis-ci joins09:41
fcrepo4/fcrepo4#3562 (BNodesAreBNodes - b1738b9 : A. Soroka): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/9afaa0d532be...b1738b9f3f76
Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/58439795
* travis-ci leaves
<f4jenkins>Yippee, build fixed!09:46
Project fcrepo-message-consumer build #1113: FIXED in 4 min 42 sec: http://jenkins.fcrepo.org/job/fcrepo-message-consumer/1113/
* github-ff joins09:49
[migration-utils] mikedurbin pushed 1 new commit to master: http://git.io/vvlpl
migration-utils/master 445a679 Daniel Lamb: Added support for datastream properties.
* github-ff leaves
* github-ff joins09:50
[migration-utils] mikedurbin closed pull request #5: Datastream properties (master...datastream-properties) http://git.io/vvlhv
* github-ff leaves
* github-ff joins09:51
[fcrepo4] ajs6f pushed 1 new commit to BNodesAreBNodes: http://git.io/vvlh0
fcrepo4/BNodesAreBNodes 41ff997 ajs6f: Removing deskolemization cruft
* github-ff leaves
* dwilcox leaves09:52
* travis-ci joins09:53
fcrepo4-labs/migration-utils#11 (master - 445a679 : Daniel Lamb): The build passed.
Change view : https://github.com/fcrepo4-labs/migration-utils/compare/cbf42f8bf181...445a67972339
Build details : http://travis-ci.org/fcrepo4-labs/migration-utils/builds/58443975
* travis-ci leaves
* travis-ci joins10:08
fcrepo4/fcrepo4#3563 (BNodesAreBNodes - 41ff997 : ajs6f): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/b1738b9f3f76...41ff997d5536
Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/58444194
* travis-ci leaves
<f4jenkins>Project fcrepo4-release-tests build #171: UNSTABLE in 1 min 22 sec: http://jenkins.fcrepo.org/job/fcrepo4-release-tests/171/10:14
* beermintor joins10:21
* tecoripa joins10:28
<sprater>quick question about fixity and checksums:
any way to override/configure the algorithm used to generate checksums?10:29
i.e., SHA256 instead of SHA1?
I remember we talked about this a long time ago...
but I don't find, nor recall, what the outcome was10:30
(btw: talkin' Fedora 4)
<escowles>sprater: IIRC, sha1 is required by modeshape -- and it would be a lot of work to switch to a different algorithm10:39
* dwilcox joins10:41
<awoods>sprater: what escowles said
* dwilcox leaves10:42
* dwilcox joins
<escowles>awoods: acoburn: my standup report: I'm working on adding an option to use the fcrepo-webapp-plus-rbacl.war instead of the unauthenticated war file, to demonstrate that audit events would pickup the username correctly10:43
<awoods>escowles: feel free to create a ticket
<escowles>awoods: https://jira.duraspace.org/browse/FCREPO-146210:44
<ajs6f>sprater: It would be, I think, a question of separating the use MODE makes of the calculation and the use you might be making of it. MODE isn't really interested in fixity as much as it is interested in hashing.
<sprater>escowles,awoods: that's what I thought... thanks for confirming.
<awoods>[standup]: reviewing/testing blanknodes PR and other PRs that float by.
<sprater>ajs6f: so that would me adding our own property, in addition to the the fcr:fixity one created by MODE?10:45
<acoburn>[standup] I'm working on FCREPO-1452. It's mostly ready but I still need to move the log4j stuff out
<escowles>ajs6f: sprater: it might be easier to update modeshape to calculate other checksums in addition to sha1, as long as it was reading the data and calculating the sha1
<ajs6f>sprater: Could be.
escowles: Maybe. I don't know that I want to couple any more tightly than we now do to MODE. I'd kind of rather do an independent fixity service that could be spec'd properly and implemented over some other platform.10:46
(e.g. Marmotta)
<sprater>escowles,ajs6f: what ajs6f said. That's more what I had in mind. There's fixity for modeshape internal purposes (I saw that it was part of the filesystemconnector), and then there's fixity for other purposes.10:47
<awoods>sprater: it looks like Peter took the tech-list approach for the same question...10:50
<ajs6f>One of my favorite examples here is large video files. Our A/V archivist has told me on occasion that the kinds of checksums we normally do just aren't very useful to him. Telling him that a fifty-gig video file "has changed. Somehow. At least one bit has changed. Somewhere." is just not that useful. He's got lots of great ideas about what kinds of more sophisticated notions of fixity might be useful in such cases. None of them are goin
<sprater>awoods: yeah, we're approaching it from both ends.10:51
<awoods>sprater: UWisc SHA sandwich10:52
<sprater>awoods: UWisc: Eating Fedora for lunch.10:53
<ajs6f>UWisc: Suffering from bad indigestion.10:57
osmandin: I'm totally confused by your problems with FCREpo-1411. It seems to me that if a triple has a subject that is the URI of the request in progress, a hash-URI based on that UR, or a blank node, it's good. Otherwise it is not.11:11
<osmandin>ajs6f: my problem is with testRoundTripReplaceGraph()11:12
https://github.com/fcrepo4/fcrepo4/blob/b94778562793ea0163a6f7aeab66d6544766b08b/fcrepo-http-api/src/test/java/org/fcrepo/integration/http/api/FedoraLdpIT.java#L78011:13
<ajs6f>osmandin: Yes, I don't understand what the problem is.
<osmandin>ajs6f: when that request body is converted into a model, it has the discussed F4 as a subject11:14
<ajs6f>osmandin: Right. You'll need to filter those.
<osmandin>ajs6f: unlike the IT tests you added11:15
ajs6f: so, yes, that's my question
<ajs6f>osmandin: What's your question?
<osmandin>ajs6f: what else needs to be filtered
ajs6f: there's w3 or purl as well
<ajs6f>osmandin: You've got it wrong way wrong. You don't know what needs to be filtered. You know what needs to be _kept_.
osmandin: Or to put it more simply: filter everything that is not a triple with the subject of the request, a hash-URI based on that triple, or a blnk node. Everything else goes.11:16
osmandin: Does that makes sense? Does that help you write a filter?11:18
<osmandin>ajs6f: I think I'll just push up what I have to convey the idea.11:19
<ajs6f>osmandin: Okay, that's a good way to move forward.
afk11:21
* ajs6f leaves
* sprater leaves11:41
* dwilcox leaves
* dwilcox joins11:45
* github-ff joins11:53
[fcrepo4] ajs6f pushed 1 new commit to BNodesAreBNodes: http://git.io/vv4JG
fcrepo4/BNodesAreBNodes ffcb393 A. Soroka: Update ContentExposingResource.java
* github-ff leaves
* ajs6f joins11:55
* github-ff joins11:59
[fcrepo4] osmandin opened pull request #768: 1411 (master...1411) http://git.io/vv4kJ
* github-ff leaves
* acoburn leaves12:08
* github-ff joins12:09
[fcrepo4] osmandin closed pull request #768: 1411 (master...1411) http://git.io/vv4kJ
* github-ff leaves
<ajs6f>osmandin: Why did you close that PR?
<osmandin>ajs6f: It seemed that you were done reviewing, right?12:10
<ajs6f>osmandin: Nope.
osmandin: Although it looks like I can still add comments, so no matter....12:11
* travis-ci joins12:14
fcrepo4/fcrepo4#3565 (BNodesAreBNodes - ffcb393 : A. Soroka): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/41ff997d5536...ffcb39304955
Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/58464812
* travis-ci leaves
<f4jenkins>Yippee, build fixed!12:17
Project fcrepo4-release-tests build #172: FIXED in 1 min 27 sec: http://jenkins.fcrepo.org/job/fcrepo4-release-tests/172/
<ajs6f>osmandin: Okay, that's all of my commenting. But I still feel like we're missing something. Let me try to say this again: there is no way that the round-trip test is going to work as it stands. it will have to be altered to use a filter. That's not because it was wrong before, it's because we changed the behavior of the system and therefore it is reasonable to change the tests that check that behavior. The filter will select only those 12:20
So— where are we still miscommunicating… ?
<osmandin>ajs6f: yes, that's what I wanted a confirmation on. That is, whether the test was in need of some change or the underlying new logic.12:21
<ajs6f>osmandin: Oh, okay. So this wasn't about the specifics of the change, but more about whether it is legit?12:22
<osmandin>ajs6f: yes. Sorry, I should've been more clear.
<ajs6f>osmandin: No, not your fault at all.
awoods/osmandin: Osmandin has raised a point (not his original point, just something that has come up in the conversation) that is worth thinking over. After this change (fcrepo-1411) it will not be possible to retrieve a representation of a resource, then immediately PUT it back. A client will have to filter out the triples that aren't about that resource. That's a slightly different behavior than what's there now, and we might need to12:24
That's happening mostly because we create _all kinds_ of triples that aren't about the resource and add them to represetnations, like triples about namespaces and so forth.12:25
<awoods>ajs6f: currently we can do this round-tripping because the non-repo resources are simply ignored?12:29
<ajs6f>awoods: That's half it. The other half is that repo-but-non-local resources get recorded (the very thing that osmandin is changing).12:30
* dwilcox leaves12:31
<awoods>ajs6f: This also relates to the default response to a GET. Using Prefer headers, I suspect we could round-trip a certain response representation of a resource, no?
<ajs6f>awoods: If we can elide SMT, then yes.12:32
<osmandin>afk
<awoods>ajs6f: I guess I need an example. Are we talking about SMTs? or repo-but-non-local resources, such as: <http://fedora.info/definitions/v4/repository#jcr/xml> rdfs:label "jcr/xml"^^<http://www.w3.org/2001/XMLSchema#string>12:34
<ajs6f>awoods: Both.
<awoods>ajs6f: How are SMTs an issue here?
<ajs6f>awoods: after osmandin's work, _either_ will throw an exception.
awoods: Because SMTs do not have the right subject.
<awoods>ajs6f: I think of SMTs as thinks like "creationDate" "lastModified", "primaryType",etc12:35
ajs6f: which SMTs do not have the right subject?
<ajs6f>awoods: Triples describing namespaces, for example.12:36
* dwilcox joins
<awoods>ajs6f: I am not seeing such triples when I am performing a GET on a resource12:37
ajs6f: What should I do to see them?
<ajs6f>awoods: It's possible that we got rid of them at some point, but I don't think so. They arise from:12:38
https://github.com/fcrepo4/fcrepo4/blob/master/fcrepo-kernel-impl/src/main/java/org/fcrepo/kernel/impl/rdf/impl/NamespaceRdfContext.java
awoods: In any event, if, using Prefer, we can "tailor" the response to include just triples with the "right" subject (and it seems to me that we can, or will be able to so do shortly) then we should, as you say, be able to "round-trip". I don't think this is a problem, in and of itself. I do think we might want to have a technical note about it somewhere, although I admit, it's not clear to me where would be appropriate. Maybe in the A12:43
?
<awoods>ajs6f: There are currently round-trip integration tests. Do those still pass?12:44
<ajs6f>awoods: That was the discussion between osmandin and myself. No, at least one won't, because it's not doing anything with the Prefer headers.12:45
<awoods>ajs6f: I am thinking about the tests in the module: fcrepo-integration-rdf12:46
<ajs6f>awoods: I don't know about those. We'd have to ask osmandin.
<awoods>ajs6f: it seems like we are moving in the right direction with scoping requests to the target resource. We just need to follow the same rules on the response side.12:49
<ajs6f>awoods: Agreed. This is about REST. Representations shouldn't acquire extra semantics based solely on their position in an exchange. They sometimes do, but we should try to avoid it.12:50
* dwilcox leaves13:04
* acoburn joins13:13
<osmandin>awoods: ajs6f: Catching up the conversation. Yes, 5 or so tests in fcrepo-integration-rdf will also failing due to 403.13:33
<awoods>ajs6f: I think the bnodes PR is good to go... do you have a surgical clustering of commits you would like to see for the squash? or should I just do a mega-squash?
<ajs6f>awoods: Totally up to you. I don't think the internal history of the thing is worth preserving.13:34
osmandin/awoods: Perhaps osmandin could try using Prefer headers to make them pass. For those that can't pass this way, we have new subtickets to expand the capabilities of Prefer headers. My suspicion is that they will turn out to be thing we already want to do anyway.13:35
<awoods>ajs6f +1
<ajs6f>I think we should dispense with the entire API as it now stands and replace it with one URL and an incredible world of Prefer headers.13:36
<osmandin>ajs6f: So, there needs to be a filter now that works with Prefer in FedoraLdp @GET?13:41
<ajs6f>osmandin: There already is an apparatus for that purpose in ContentExposingResource.13:42
<osmandin>ajs6f: ok, I see it. Thx.13:44
<ajs6f>kool
* github-ff joins13:56
[fcrepo4] awoods pushed 1 new commit to master: http://git.io/vvBUy
fcrepo4/master ceacc98 ajs6f: Re-Work true blank-node implementation...
* github-ff leaves
* github-ff joins
[fcrepo4] awoods closed pull request #748: BNodes are BNodes and should be treated as such (master...BNodesAreBNodes) http://git.io/p7NB
* github-ff leaves
* ajs6f1 joins14:05
* ajs6f leaves
<f4jenkins>Project fcrepo4-oaiprovider build #194: FAILURE in 21 sec: http://jenkins.fcrepo.org/job/fcrepo4-oaiprovider/194/14:09
* travis-ci joins14:13
fcrepo4/fcrepo4#3568 (master - ceacc98 : ajs6f): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/6e7e22e52822...ceacc9832890
Build details : http://travis-ci.org/fcrepo4/fcrepo4/builds/58482886
* travis-ci leaves
* osmandin leaves14:28
* osmandin joins14:34
* github-ff joins14:36
[fcrepo4-oaiprovider] awoods pushed 1 new commit to master: http://git.io/vvB0d
fcrepo4-oaiprovider/master 03c2bef Andrew Woods: Propagating impact from blank node patch...
* github-ff leaves
<f4jenkins>Yippee, build fixed!14:37
Project fcrepo4-oaiprovider build #195: FIXED in 1 min 47 sec: http://jenkins.fcrepo.org/job/fcrepo4-oaiprovider/195/
awoods: Propagating impact from blank node patch
<ajs6f1>awoods: Take a look at osmandin's latest comment to his ticket. Does that last triple seem weird to you?15:25
* Nianli joins15:44
* dhlamb leaves16:03
* osmandin leaves16:33
* Nianli leaves16:43
* whikloj leaves16:46
* ajs6f1 leaves16:51
* mikeAtUVa leaves17:00
* acoburn leaves17:05
* MohamedAR joins17:13
* jgpawletko leaves17:21
* jgpawletko joins17:49
* ksclarke leaves18:00
* jgpawletko leaves18:36
* MohamedAR leaves19:22
* dwilcox joins19:49
* MohamedAR joins19:54
* dwilcox_ joins20:02
* dwilcox leaves
* MohamedAR leaves20:09
* ajwagner leaves20:14
* ajwagner joins20:16
* MohamedAR joins20:20
* MohamedAR leaves20:29
* ksclarke joins20:59
* ksclarke leaves21:03
* ksclarke joins21:18
* jgpawletko joins21:46
* jgpawletko leaves21:49
* dhlamb joins22:11
* ksclarke leaves22:42
* dhlamb leaves23:50

Generated by Sualtam