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

Using timezone: Eastern Standard Time
* ksclarke leaves00:30
* dwilcox joins01:27
* dwilcox leaves01:32
* dwilcox joins02:28
* dwilcox leaves02:32
* dwilcox joins03:28
* dwilcox leaves03:32
* dwilcox joins04:29
* dwilcox leaves04:33
* dwilcox joins05:29
* dwilcox leaves05:34
* daines1 joins06:57
* daines leaves06:59
* dwilcox joins08:14
* dwilcox leaves08:46
* dwilcox joins08:56
* tecoripa joins09:02
* ajs6f joins09:17
* dwilcox leaves09:32
* daines1 leaves09:34
* can-of-bees joins09:38
* ksclarke joins09:45
* dwilcox joins09:51
* fasseg joins09:53
* ajs6f leaves09:54
* ajs6f joins09:58
* scossu joins10:08
* daines joins10:10
* ksclarke leaves10:12
* ksclarke joins10:19
<ajs6f>scossu: I'm starting to read your work, and finding it a little tough going: why did you add LDP -related information in the node definitions, and why did you add the indexing namespace and types?10:21
Actually, there's a lot of stuff here that seems out of place. We shouldn't have node types that concern versionability or hierarchy or exportability or RBAC or any of this other stuff— none of these properties are actually persisted ever: they are calculated.10:22
<scossu>ajs6f: I added all those properties because with a restrictive node type they have to be declared explicitly. I'm not sure if there is a better way to do that.10:24
<ajs6f>scossu: This is my point. They should _not_ have to be declared, because they don't exist in the JCR. They are constructed as needed during API calls.
Did you find that tests failed without them?
(Because they shouldn't, and that would represent some bugs to fix.)10:25
<scossu>ajs6f: Some tests were failing because properties were being added. I just defined all the properties that I could discover on a node because I didn't know which ones are actually added.10:26
<ajs6f>scossu: Okay, we may need to back up to that point. Tests should not fail if these kinds of properties are not defined, because we do not store them in the JCR.10:27
<scossu>So if you say that LDP and other Fedora default properties are calculated, I can try to remove them and re-run the tests.
<ajs6f>scossu: Yes, that might be a good move. Then we can fix those problems before moving on rearrange the "legit" types.10:28
<scossu>ajs6f: Ok, I will do that. I actually was planning on going back and pruning unneeded definitions but only when all tests passed - unless those explicit definitions are messing up something else.10:30
<ajs6f>scossu: There's another thing about this: it looks like you are constructing your test data "directly" (using the JCR APi). If you are going to assert your conditions using HTTP (and you are), you probably want to construct your test data using the HTTP API as well, just to take that variable out of the picture.10:31
scossu: I don't know whether they are messing anything up, but I do know 1) They shouldn't be needed and if they are, that's a bug, and 2) They make it much more difficult to understand what is going on here. So in order to get some clarity and simplify the problem, we should prune. You'll have to chalk it up to my old, slow brain. :)10:32
<scossu>ajs6f: Ok, so I'll remove fedora:hierarchyNode, fedora:versionable, fedora:exportable, fedora:rbac and see how it goes. Anything else?10:34
<ajs6f>scossu: Let me see… I had a diff p here a sec ago.
<scossu>Oh, and ldp:resource.
fedora:datastream seems to have new properties like "image:metadata (nt:unstructured) multiple", but maybe that wasn't you?10:36
<scossu>ajs6f: re: test data: I was following the other test examples I saw in the same class. Most of them I used the HTTP API by defining an addMixin method in AbstractResourceIT.10:37
<ajs6f>Yes, but you aren't, you're doing:
and before that, you are creating impl-typed obejcts directly, e.g.
new FedoraResourceImpl
Definitely use the *Service classes to directly create things, but even better to use the HTTP API. Let me find you a good exampe.10:38
(And yes, I realize this stuff is horribly undocumented. :) )
There's a nice short example.
<scossu>ajs6f: that's definitely a mistake. I will change that to use the API method.10:40
<ajs6f>scossu: Rocking. In theory, it should have no effect, but I really want to cut this down to your actual problem.
<pivotal-bot____>A. "Torcrexilly" Soroka added "Document practices for integration testing" https://www.pivotaltracker.com/story/show/6651050610:42
<scossu>ajs6f: Ok, thanks and sorry about the clumsiness. This is the first real Java project I'm dealing with...
<ajs6f>scossu: Then you are doing amazingly well and should be proud of yourself so far.
<scossu>I'll get back when I clean up and re-run the tests.
<ajs6f>scossu: Groovy. Feel free to shout out for help along the way.10:43
<pivotal-bot____>A. "Torcrexilly" Soroka edited "Move tests of classes which were moved kernel-api" https://www.pivotaltracker.com/story/show/6645471210:47
<ajs6f>Standup at 11 ET at: • https://plus.google.com/hangouts/_/event/c1glu6soq43r1rr6ou17qtobug810:48
* gregjansen joins10:54
<pivotal-bot____>Gregory Jansen edited "Establish performance goals for UNC cluster" https://www.pivotaltracker.com/story/show/6550482411:08
* tecoripa leaves11:14
* gregjansen leaves11:21
* gregjansen joins11:23
<pivotal-bot____>Frank Asseg added "Add SPARQL actions to bench-tool" https://www.pivotaltracker.com/story/show/66514236
<ajs6f>all: Add agenda items for tomorrow's committers' call to:
fasseg: Is this RADAR?11:27
benpennel2: Are you prepping a PR for the benchtool as the outcome of this work?11:33
* tecoripa joins11:41
* gregjansen leaves11:54
* gregjansen joins11:55
* scossu leaves12:03
<fasseg>ajs6f: indeed this is RADAR
<ajs6f>fasseg: Cool. It makes me happy to put links in the wiki. :)12:04
* scossu joins12:06
* dwilcox leaves
* dwilcox joins12:07
* dwilcox leaves12:11
* dwilcox joins12:17
* tecoripa leaves12:26
* dwilcox leaves12:41
* dwilcox joins13:15
* edInCo joins
* dwilcox leaves13:20
* benpennell joins14:40
* tecoripa1 joins14:41
* CanOfBees joins
* tecoripa1 leaves
* ksclarke leaves14:47
* can-of-bees leaves
* benpennell2 leaves
* scossu leaves14:56
* ajs6f leaves14:59
* ajs6f joins15:01
* ksclarke joins15:04
* ajs6f1 joins15:14
* ajs6f leaves
* CoB joins15:15
* ajs6f1 leaves
* ajs6f joins
* gregjansen leaves15:16
* CanOfBees leaves15:18
* ajs6f1 joins15:21
* ajs6f leaves
* scossu joins15:22
* ajs6f joins15:27
* ajs6f1 leaves
* Guest64776 leaves15:40
* dwilcox joins15:42
* ajs6f leaves15:49
* ajs6f joins15:51
* ajs6f leaves
* ajs6f joins15:52
* edInCo leaves15:59
* edInCo joins16:01
<pivotal-bot____>Benjamin Pennell added comment: "https://github.com/futures/benchtool/pull/10" https://www.pivotaltracker.com/story/show/6634383216:15
Benjamin Pennell finished "Modify benchtool to support testing transactions" https://www.pivotaltracker.com/story/show/66343832
<ajs6f>fasseg: can you review benpennel's PR on bench tool?
* gregjansen joins
<benpennell>i've done some already, but i'm thinking i should make another ticket for getting stats comparing performance with and without transactions16:16
<ajs6f>benpennel: +1
Good to keep track of these things.
<benpennell>also this isn't tested against a cluster yet since i don't have one available and ready, so while it should probably be okay i can't say for sure16:17
<ajs6f>benpennell: That's fine. We know that we already have some issues with xactions against clusters (we don't yet have a notion of "global transaction") so that's okay.
If gregjansen is able to get the UNC cluster stable, certainly you can try there.16:18
<pivotal-bot____>Benjamin Pennell added "Benchmark performance of using transactions versus no transactions" https://www.pivotaltracker.com/story/show/6654432816:20
Benjamin Pennell started "AccessRoles session logout when post of new roles fails validation" https://www.pivotaltracker.com/story/show/6598005816:21
* ajs6f1 joins16:23
* benpennell1 joins16:27
<gregjansen>so I am seeing a leveldb lock, just now showing up now that I am using the ModeShapeEngine instead of JcrRepositoryFactory to deploy the repo..
IO error: lock /home/count0/fcrepo4/fcrepo-transform/target/data/LOCK: already held by process
anyone remember that one? Only getting it for fcrepo-transform and not for other ITs preceding those. I have told failsafe to not reuse forks16:28
<ajs6f1>I don't remember it specifically, but I wonder if switcing away from leveldb might get rid of it for the moemnt.
* ajs6f leaves16:29
<ajs6f1>Might take a look at the MODE JSON config or the ISPN config in use for those i-tests. They might be different somehow?
<gregjansen>probably.. failsafe may still have been creating a fork actually.. looks like default forkCount is 116:31
<ajs6f1>Hm. Make that zero? I think forkMode is deprecated.16:32
* benpennell leaves16:34
<pivotal-bot____>Benjamin Pennell started "AccessRolesTypes refactor to remove static flag checks and cleanup" https://www.pivotaltracker.com/story/show/6597974417:00
* gregjansen leaves17:04
* gregjansen joins17:06
ajs6f1: yeah I made it zero. then when that didn't work I removed leveldb from that IT config. Now making progress again.17:11
I'm out for the day. Good luck recreating the momentum!
* ajs6f1 leaves
* gregjansen is whistling in the dark17:13
* fcrepo-bot joins17:47
* dwilcox leaves17:49
* dwilcox joins
* dwilcox leaves17:54
* edInCo leaves18:19
* ksclarke leaves18:23
* fcrepo-bot leaves18:52
* scossu leaves19:04
* fcrepo-bot joins19:10
* fasseg leaves19:55
* fcrepo-bot leaves20:19
* scossu joins22:40
* scossu leaves23:46
* ksclarke joins00:33
* ksclarke leaves00:41
* ksclarke joins00:43

Generated by Sualtam