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

Using timezone: Eastern Standard Time
* nbanks joins01:35
* kaarefc joins04:13
* fasseg joins05:52
* unai_ joins06:46
<fasseg>some profiling results are at: https://wiki.duraspace.org/pages/viewpage.action?pageId=3465367307:05
Hey Unai welcome to #duraspace-ff07:06
nbanks eddies: have you guys seen org.fcrepo.utils.LowLevelCacheEntryTest.testFileSystemExternalIdentifier() fail with a java.lang.NoSuchFieldException: blobCacheName before?07:15
when unai_ and leenatat are trying to build fcrepo they get that one, and I have no idea where this comes from
unai_: did you do a git pull before building today?
<eddies>fasseg: that sounds not dissimilar from the most recent posting to ff-tech07:16
same test class at least
what platforms are unai_ and leenatat using?
<fasseg>win
<unai_>Hello everyone!
<eddies>i wonder if that's the issue
<unai_>Yes, windows 7
<eddies>i don't think any of the other committers are using windows so it wouldn't surprise me if there's some platform-specific issue07:17
<fasseg>oh and can you add leenata.bankhele@gmail.com and goni.otegi.unai@gmail.com to the google groups?
<eddies>they should each make a subscribe request07:18
<fasseg>yeah this seems to be exactly the same problem: https://groups.google.com/forum/?hl=en&fromgroups#!topic/ff-tech/uWhBZfSPc3E
<eddies>well, there's a first ticket for unai_ ;-)07:19
welcome aboard! =)
<fasseg>ok so unai_ pls send requests for https://groups.google.com/forum/?hl=en&fromgroups#!forum/ff-tech and https://groups.google.com/forum/?hl=en&fromgroups#!forum/ff-internal
<unai_>:)
Ok, good start
<fasseg>eddies: pivotal accounts are 'goniotegiunai' and 'Leenata'07:20
<eddies>fasseg: can you add them, please? you are an owner of the project too07:21
same for github07:22
<fasseg>opps didnt know...will do
* leenata joins
<unai_>Requests for both groups sent07:23
<eddies>unai_ i don't see any request for https://groups.google.com/forum/#!forum/ff-internal07:26
<unai_>The only option that I have is to send a message07:27
<fasseg>hey leenata!07:32
* github-ff joins
[fcrepo4] eddies pushed 2 new commits to master: http://git.io/oTCmjA
fcrepo4/master 08e32c0 Edwin Shin: Removes old Velocity templates from http-api
fcrepo4/master 259ab4e Edwin Shin: Removes hard-coded "/rest/" from @Path annotations. The path "/rest" is now set in web.xml (and so is technically variable)....
* github-ff leaves
<fasseg>To reiterate for leenata: Nothing is known about the exception you guys get on your windows boxes and there is a post on the techie list with the same issue https://groups.google.com/forum/?hl=en&fromgroups#!topic/ff-tech/uWhBZfSPc3E07:34
So your current task is to investigate the test failure and try to get fcrepo4 running on linux
err * on windows07:35
<leenata>hey07:36
<fasseg>eddies: have you taken a look at the results I posted at https://wiki.duraspace.org/pages/viewpage.action?pageId=34653673?
<leenata>yes.. we are trying to figure that out
<fasseg>there def is something slow in the lucene hanlding but theres also a reflection call in modeshape which seems too long for me
and this is still only creating an empty object...07:37
so lucene should do almost nothing
leenata: nice
and feel free to post comments on ff-tech for this issue...;)07:38
<unai_>Ok, we will say something07:39
;)
<bljenkins>Project fcrepo-fixity-corrupter build #51: STILL FAILING in 46 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity-corrupter/51/07:48
Project fcrepo-fixity build #226: UNSTABLE in 3 min 49 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/226/07:52
* travis-ci joins
[travis-ci] futures/fcrepo4#639 (master - 259ab4e : Edwin Shin): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/ff4accf91567...259ab4e799e5
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/7802633
* travis-ci leaves
<eddies>fasseg: could you do a sanity-check of the create object profiling on a vanilla instance of modeshape?07:56
i'd like to be sure it's really isolated to modeshape and not something we're doing either in our branch of modeshape ir something in how we're wiring up modeshape in fcrepo4
(i'd say just build modeshape master yourself or grab a 3.2 binary if you'd really rather—although master is closer to what we're currently using in fcrepo4)07:57
<fasseg>ok im updating the fixity-corrupter atm. So ill finish this and then profile modeshape, ok?07:58
<eddies>ok
fyi, if you build modeshape yourself, give yourself a lot of permgen, i've been using MAVEN_OPTS="-Xmx512m -XX:MaxPermSize=256m"07:59
and it's mvn install -s settings.xml (i would also just add -DskipTests to speed things up)08:00
<fasseg>okay08:01
these CATALINA_OPTS helped me a lot against permgens btw: -Djava.awt.headless=true -Dfile.encoding=UTF-8 -server -Xms1536m -Xmx1536m -XX:NewSize=256m -XX:MaxNewSize=256m -XX:PermSize =256m -XX:MaxPermSize=256m -XX:+DisableExplicitGC08:02
grml custom settings.xml in a project...08:03
that desintegrated my local repo once, because someone had overwritten all old versions of jars with newer ones in a project repo08:04
<cbeer_>eddies, fasseg: i'd like to remind us that the MODE indexing is configurable to be sync or async (and when it's turned off, creating objects is that much faster)... so getting bogged down looking at that is probably not a great use of time, no?
<fasseg>are we currently in sync mode?08:05
then ill just rerun the tests with a diff configuration first
but we still need the modeshape indexing for the fieldsearch don't we?08:06
<cbeer_>yes, sync is the default.
and we only need sync if we want the field search to be absolutely consistent. when we had this discussion last week, it wasn't clear that that was actually important.08:07
* unai_ leaves
* unai_ joins08:11
<eddies>cbeer: sync is only important when it's really important :P08:31
we had this issue with the resourceindex queries back in the early days too08:32
that is, we had async by default, but then folks were building application workflows where they needed to immediately query for some triple that they had just added08:33
<fasseg>we could emit a message if an object was indexed so that client's now when they can query objects08:35
*clients know
<eddies>and yes, i know that MODE supports async—i was just pointing awoods at the configuration for that a couple days ago. i just want the sanity check of knowing if the creation times we're seeing and the apparent time spent in lucene is the same in vanilla MODE to establish a baseline before we embark on other optimizations08:36
<fasseg>or a header value for the http request? something like 'Fedora-Index: Async'
<eddies>i think it's more likely we end up doing async by default. and then look into what workarounds/process we offer to clients who need sync behavior.08:37
but i still want the MODE baseline established
<fasseg>kk08:38
<bljenkins>Yippie, build fixed!08:47
Project fcrepo-fixity-corrupter build #52: FIXED in 1 min 37 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity-corrupter/52/
* github-ff joins08:53
[fcrepo4] eddies pushed 1 new commit to master: http://git.io/FUE1lg
fcrepo4/master 92afb25 Edwin Shin: Adds a stub for an index.html page
* github-ff leaves
<bljenkins>Project fcrepo-kitchen-sink build #344: UNSTABLE in 7 min 35 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-kitchen-sink/344/08:55
<fasseg>^^ the fixity corrupter globbed a bit too much path09:02
<bljenkins>Project fcrepo-kitchen-sink build #345: FAILURE in 19 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-kitchen-sink/345/09:04
* nbanks leaves09:05
<bljenkins>Project fcrepo-fixity-corrupter build #53: SUCCESS in 1 min 9 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity-corrupter/53/09:06
* jonathangee leaves09:09
* travis-ci joins09:11
[travis-ci] futures/fcrepo4#640 (master - 92afb25 : Edwin Shin): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/259ab4e799e5...92afb2517ba5
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/7804741
* travis-ci leaves
* gregjansen joins09:15
<bljenkins>Project fcrepo-fixity-corrupter build #54: SUCCESS in 1 min 3 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity-corrupter/54/09:16
* jonathangee joins09:17
* unai_ leaves
<awoods>Regarding performance optimizations, now that we have re-established a baseline for Fedora4 in single-server mode, it may be worth evaluating the comparison with 5? 10? server clusters.09:19
<bljenkins>Project fcrepo-fixity build #227: STILL UNSTABLE in 4 min 50 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/227/
Project fcrepo-kitchen-sink build #346: NOW UNSTABLE in 4 min 59 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-kitchen-sink/346/09:21
Project fcrepo-fixity build #228: STILL UNSTABLE in 3 min 45 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/228/09:31
* unai_ joins
* ff-github joins09:34
[fcrepo-fixity] fasseg pushed 1 new commit to master: http://git.io/lZ3Tkg
fcrepo-fixity/master 6ea778d fasseg: changed test db to a in memory hsqldb instead of a file db
* ff-github leaves
* ff-github joins09:45
[fcrepo-fixity] fasseg pushed 1 new commit to master: http://git.io/_RgCoA
fcrepo-fixity/master f8efb62 fasseg: made some changes to adhere to the sonar rules
* ff-github leaves
* unai_ leaves09:48
* unai_ joins
<bljenkins>Project fcrepo-fixity build #229: STILL UNSTABLE in 4 min 40 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/229/09:49
<fasseg>after the standup ok?09:52
<awoods>fasseg: sure09:56
* ajs6f joins09:58
* unai_ leaves
* ff-github joins10:07
[fcrepo-fixity] fasseg pushed 1 new commit to master: http://git.io/-EOoZA
fcrepo-fixity/master cf3cf46 fasseg: removed some more sonar violations
* ff-github leaves
<pivotal-bot_>Andrew Woods added "Establish performance baseline against cluster configuration" https://www.pivotaltracker.com/story/show/5115190310:09
Andrew Woods edited "Establish performance baseline against cluster configuration" https://www.pivotaltracker.com/story/show/51151903
* github-ff joins10:10
[fcrepo-kitchen-sink] eddies pushed 1 new commit to master: http://git.io/XCl-_Q
fcrepo-kitchen-sink/master f6c8c7f Edwin Shin: Updates web.xml servlet-mapping for jersey-servlet to match fcrepo-webapp....
* github-ff leaves
<bljenkins>Project fcrepo-fixity build #230: STILL UNSTABLE in 5 min 7 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/230/10:11
<pivotal-bot_>Edwin Shin accepted "Have Maven Checkstyles" https://www.pivotaltracker.com/story/show/4983697910:12
Edwin Shin accepted "Update JMeter test plan to work against current REST API" https://www.pivotaltracker.com/story/show/50008127
Edwin Shin accepted "Archives Sequencer build is broken" https://www.pivotaltracker.com/story/show/50462891
Edwin Shin accepted "Support conditional PUT requests" https://www.pivotaltracker.com/story/show/4967150510:13
Edwin Shin accepted "Extract predicate literals into a centralized location in fcrepo-kernel (for ease of changing when we know what the..." https://www.pivotaltracker.com/story/show/50856301
* ff-github joins10:14
[fcrepo-fixity] fasseg pushed 1 new commit to master: http://git.io/EOHPiQ
fcrepo-fixity/master 63c2cf0 fasseg: closing resources in exception
* ff-github leaves
<bljenkins>Yippie, build fixed!10:15
Project fcrepo-kitchen-sink build #347: FIXED in 5 min 19 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-kitchen-sink/347/
Project fcrepo-fixity build #231: STILL UNSTABLE in 4 min 2 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/231/10:18
<pivotal-bot_>Andrew Woods added comment: "Are you ready for a pull-request, Adam?" https://www.pivotaltracker.com/story/show/4920564910:19
* github-ff joins
[fcrepo4] eddies pushed 1 new commit to master: http://git.io/uS8F4A
fcrepo4/master 9b63f43 Edwin Shin: Hey, look! A favicon!
* github-ff leaves
* barmintor joins10:20
<pivotal-bot_>A. "Codiselic" Soroka added comment: "This was taken over a good while ago by nbanks, who incorporated it into an issue that he was/is w..." https://www.pivotaltracker.com/story/show/49205649
Edwin Shin added comment: "I cropped a Fedora image to make a favicon in fcrepo-webapp/src/webapp/favicon.ico. Provided the Fedora webap..." https://www.pivotaltracker.com/story/show/5014472910:21
Edwin Shin finished "Create favicon.ico for fcrepo4" https://www.pivotaltracker.com/story/show/50144729
<awoods>ajs6f: So maybe 49205649's state should be put back to "started" and the Owner changed to Nigel?10:22
<ajs6f>Sure, if that's our workflow. I'm fine with that.
I'll do that now.
<awoods>You da man. That makes sense to me.10:23
<pivotal-bot_>A. "Codiselic" Soroka started "Inject authenticated sessions into the JAX-RS resources" https://www.pivotaltracker.com/story/show/49205649
A. "Codiselic" Soroka unstarted "Add an OAuth2 service to the REST API's JAX-RS container using Spring Security OAuth2" https://www.pivotaltracker.com/story/show/45458249
<ajs6f>Spring used to be my weird friend. Now it's turned on me and it's like walking into a buzzsaw every time I have to deal with it.10:24
<barmintor>chin up!10:25
<ajs6f>That just gives Spring the opportunity for an uppercut.
barmintor: I pushed all the contexts together. Now the filters work, but the endpoints don't get injected.10:26
{sigh}
<eddies>A. "Glass Jaw" Soroka vs. Spring "The False Friend"
<ajs6f>I'd do better against Sonny Liston.
<barmintor>the endpoints… don't get… injected...
<ajs6f>Right. With a SessionFactory.10:27
<barmintor>the jaxrs component scan is not working?
oh, the sessionfactory
<ajs6f>Yeah, the scan seems to be working, but it blows out trying to init the endpoints.
<barmintor>they don't get @Inject'ed, or they don't get @Autowire'ed ?
<ajs6f>@Autowire. It's in AbstractResource.10:28
<barmintor>Hmm
<ajs6f>(Which the endpoints subclass.)
<fasseg>hmpf sonar thinks that jena Statements are closeable and it complains, do we use Supress annotaions for that?
<ajs6f>Spring! Just inject my stuff with things!
If Statements are really closeable we should just use try-with-resources. Or are they not really Closeable?10:29
<cbeer_>barmintor: the steering group is even more wishy-washy about features than you are..
i wonder if we're saying we want to accomodate people who use e.g. SOAP
<fasseg>It's not closeable
<barmintor>cbeer_: I'm destined for management
<ajs6f>That's weird. I wonder how Sonar is miscalculating that. It seems straightforward.
<fasseg>at least i cant see a method, is it possible that sonar thinks this is a sql statement?
<ajs6f>It should reflect on the type hierarchy.10:30
CLoseable and Autocloseable are simple interfaces.
Tell them we don't do SOAP, but we will do detergent.
<fasseg>i would think so too, but i thought maybe someone wrote a regex for that...10:31
<ajs6f>I have no idea what Sonar is really doing. Eclipse is smart enough to get it right.
barmintor: hm. Looks like it's trying to init the providers (e.g. HtmlProvider) twice. Maybe that's a clue...10:33
<cbeer_>awoods: i read your rejection comments.. i think most of them are already reflected in other tickets and the original ticket is worth closing out (but we should double check that)
<ajs6f>barmintor: once, it works, then the second time it blows out the same way (not finding injected resources).
I wonder if it's trying to init everything twice...
I guess I can profile it and start poking through the heap to see what objects actually get built.10:34
<awoods>cbeer_: I suspect most of the issues I saw have nothing to do with your ticket... however, I was not able to test your ticket due to the thicket of errors.
<bljenkins>Project fcrepo-fixity-corrupter build #55: SUCCESS in 1 min 15 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity-corrupter/55/10:35
<pivotal-bot_>Edwin Shin added comment: "Currently, we're pushing builds of metrics from futures/metrics to our nexus repo. Once Metrics hits 3.0 fina..." https://www.pivotaltracker.com/story/show/5082740310:37
<cbeer_>eddies: i saw metrics did a rc1 release
<eddies>cbeer_: yeah, i figured we'd keep on the bleeding edge for now anyway, i just pulled in the latest upstream changes to our fcrepo-snapshot branch10:38
<cbeer_>k
* travis-ci joins10:39
[travis-ci] futures/fcrepo4#641 (master - 9b63f43 : Edwin Shin): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/92afb2517ba5...9b63f43b0147
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/7807255
* travis-ci leaves
<fasseg>StatementImpl doesn implement Closeable it's just Statement and FrontsTriple, so no idea why sonar does complain
<bljenkins>Project fcrepo-fixity build #232: STILL UNSTABLE in 4 min 44 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/232/
Project fcrepo-kitchen-sink build #348: SUCCESS in 5 min 42 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-kitchen-sink/348/10:41
<pivotal-bot_>Edwin Shin accepted "Fix sitemap URLs to use full paths to nodes" https://www.pivotaltracker.com/story/show/5100539110:42
* leenata leaves10:43
* ajs6f leaves10:44
<pivotal-bot_>Edwin Shin added comment: "See: http://fcrepo4.fcrepo.org/fcrepo/ (or http://fcrepo4.fcrepo.org/fcrepo/favicon.ico)" https://www.pivotaltracker.com/story/show/50144729
Edwin Shin accepted "Create favicon.ico for fcrepo4" https://www.pivotaltracker.com/story/show/50144729
<cbeer_>+1 to http://fcrepo4.fcrepo.org/fcrepo/10:45
<eddies>lots of folks joining ff-tech today
i think i've added half-a-dozen folks to the list today
* ff-github joins10:46
[fcrepo-fixity] fasseg pushed 1 new commit to master: http://git.io/abNvAQ
fcrepo-fixity/master 8c8b9e4 fasseg: added //NOSONAR comment for false positive
* ff-github leaves
<cbeer_>eddies: i just presented it to the hydra partners meeting
<awoods>eddies: can you add me as manager to ff-tech?
<eddies>well, it worked
i think you are already, no?
<awoods>no
ff-internal, yes
ff-tech, no
<eddies>done10:47
<awoods>cheers
<bljenkins>Project fcrepo-fixity build #233: STILL UNSTABLE in 3 min 16 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/233/10:49
* ajs6f joins
<cbeer_>barmintor: can you/did you send a public link to your slides to ff-tech?10:50
<barmintor>I wa going to share the doc to the group, but I haven't yet
one sec
<awoods>eddies: Do we have a Fedora Futures AWS account?10:52
<pivotal-bot_>Chris Beer added "java.lang.NullPointerException at org.fcrepo.syndication.RSSPublisher" https://www.pivotaltracker.com/story/show/5115527910:53
* gregjansen leaves
<pivotal-bot_>Chris Beer added comment: "Also in webhooks." https://www.pivotaltracker.com/story/show/5115527910:54
Chris Beer added comment: "http://fcrepo4.fcrepo.org/fcrepo/rest/fcr:search was fixed10:55
I think I changed the node template and broke th..." https://www.pivotaltracker.com/story/show/50741659
* nbanks joins10:57
<pivotal-bot_>Andrew Woods edited "Profile report of jmeter test plan" https://www.pivotaltracker.com/story/show/5099792910:58
* fasseg leaves11:00
* bljenkins leaves
* bljenkins joins
<eddies>barmintor: standup time11:01
<barmintor>eddies: ok!11:03
<pivotal-bot_>Andrew Woods started "Profile report of jmeter test plan" https://www.pivotaltracker.com/story/show/5099792911:12
Andrew Woods added "Compare direct modeshape object creation to fcrepo4 object creation" https://www.pivotaltracker.com/story/show/5115721511:14
Andrew Woods edited "Compare direct modeshape object creation to fcrepo4 object creation" https://www.pivotaltracker.com/story/show/51157215
Andrew Woods edited "Compare direct modeshape object creation to fcrepo4 object creation" https://www.pivotaltracker.com/story/show/5115721511:15
Andrew Woods started "Compare direct modeshape object creation to fcrepo4 object creation" https://www.pivotaltracker.com/story/show/51157215
Benjamin Armintor started "Update LowLevelStorageService to be cluster-aware" https://www.pivotaltracker.com/story/show/4773060711:16
* fasseg joins11:29
* barmintor_ joins11:40
* barmintor leaves11:41
<awoods>fedora3: https://raw.github.com/futures/ff-jmeter-testResults/master/2013-02-10-FedoraTests/run2/results/ThreadCountVsElapsedTime.png11:43
fedora4: https://wiki.duraspace.org/display/FF/Fedora+4+test+results+-+2013-06-03
* nbanks_ joins
<fasseg>barmintor, ajs6f: this is the screenshot of the profiling sessions11:45
https://wiki.duraspace.org/pages/viewpage.action?pageId=34653673
* nbanks leaves11:47
* nbanks_ leaves11:48
* bljenkins leaves
* barmintor leaves11:49
* bljenkins joins11:54
* nbanks joins11:55
ben: https://github.com/nigelgbanks/fcrepo4/tree/49205649-injectable-sessions-and-49012389-workspace11:57
* barmintor joins12:25
<pivotal-bot_>Andrew Woods added "Refactor fcrepo-rss and frcrepo-webhooks out of fcrepo4" https://www.pivotaltracker.com/story/show/5116425312:39
Andrew Woods edited "Compare direct modeshape object creation to fcrepo4 object creation" https://www.pivotaltracker.com/story/show/5115721512:40
Andrew Woods edited "Profile report of jmeter test plan" https://www.pivotaltracker.com/story/show/50997929
Andrew Woods added "Optimize indexing configuration" https://www.pivotaltracker.com/story/show/5116453912:43
Andrew Woods edited "Optimize indexing configuration" https://www.pivotaltracker.com/story/show/51164539
* github-ff joins12:48
[fcrepo4] barmintor pushed 1 new commit to distexec: http://git.io/5zYhJw
fcrepo4/distexec ae56df3 Benjamin Armintor: DRY; make sure correct externalId set for composite stores
* github-ff leaves
* ajs6f leaves
<pivotal-bot_>Andrew Woods edited "Establish performance baseline against cluster configuration" https://www.pivotaltracker.com/story/show/5115190312:55
* awoods leaves13:05
* barmintor leaves
* bljenkins leaves
* barmintor joins13:08
* bljenkins joins
* awoods joins13:12
<nbanks>ajs6f: I did a small test of custom injections via Jersey, it seem works correctly https://github.com/nigelgbanks/jittery/13:13
* jonathangee leaves13:33
<barmintor>The method thenReturn(AdvancedCache<capture#1-of ?,capture#2-of ?>) in the type OngoingStubbing<AdvancedCache<capture#1-of ?,capture#2-of ?>> is not applicable for the arguments (AdvancedCache<capture#3-of ?,capture#4-of ?>)14:08
Generics for everyone!
testing testing twsting14:24
<pivotal-bot_>Andrew Woods started "Optimize indexing configuration" https://www.pivotaltracker.com/story/show/5116453914:38
* ajs6f1 joins14:54
nbanks: I know it can. It worked in my "injection" branch. I'm beginning to suspect that something changed in the Fedora code base between then and now.14:55
<barmintor>changes?! IN THE FEDORA CODE BASE?!?!
* barmintor 's eyes twitch as the mania of testing overtakes him15:24
<ajs6f1>*'s eyes twitch as the naked horror of Spring overwhelms him*
barmintor: the only way I can figure to do this (have beans in the Jersey-controlled scanned context use all the stuff it needs) is to drag the entire repo gear into the scanned ambit. To have all of the beans scanned. So that's what I'm going to do.15:26
Scan all the things.
<barmintor>I am in favor of "Get it to work, and then get it to work correctly"15:27
<cbeer_>+1
<ajs6f1>Well, it should make eddies happy. It will cut down the XML to "scan=everything".
afk bbi1015:28
* ajs6f1 leaves
<barmintor>cbeer: I hope what you wanted from this ticket is to use ISPN's built-in cluster magic to do the fixity checks on local caches, b/c that's what I'm doing15:29
<cbeer>works for me
<barmintor>okey doke15:30
given that, I think I'm ginning up the last unit test
then I need to run this branch in a cluster :P
* kaarefc leaves15:31
* ajs6f joins15:36
Urg. That's not going to work either. URRRRGHH.15:39
There has got be a way this makes sense.
* github-ff joins15:48
[fcrepo4] barmintor pushed 1 new commit to distexec: http://git.io/LHQpsA
fcrepo4/distexec 9d294df Benjamin Armintor: Move CacheLocalTransform into a class, testing
* github-ff leaves
<cbeer>barmintor: i just want to check, but i assume the work you're doing still works with a non ISPN binary store, right?15:57
<barmintor>cbeer: falls back to prev impl
<cbeer>ok. there's just not enough context in the diff
<barmintor>cbeer: https://github.com/futures/fcrepo4/blob/9d294df6e726d895024c552e8fb12fcd585ab6fc/fcrepo-kernel/src/main/java/org/fcrepo/services/LowLevelStorageService.java#L15815:59
<cbeer>+1
<barmintor>Ooh, that javadoc is totally wrong
gotta fix that16:00
downside of this clustering stuff: we get into the "your transformation class has to be serializable" jazz16:03
oh well
<bljenkins>Yippie, build fixed!16:10
Project fcrepo4 build #751: FIXED in 22 min: http://ci.projectblacklight.org/jenkins/job/fcrepo4/751/
<barmintor>Hmm. Looks like I need a way to go from Cache to BinaryStore
<cbeer>can't.
cache is an ISPN thing
BinaryStore is a MODE thing16:11
<barmintor>ruh roh
<cbeer>why?
<bljenkins>Project fcrepo-fixity-corrupter build #56: SUCCESS in 1 min 6 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity-corrupter/56/
<barmintor>can I get the inputstream frm a cache w/o going to the binarystore?
<cbeer>yes. the cache is what provides the stream16:12
<barmintor>Hmm
LowLevelCacheEntry makes much use of the BinaryStore
opk16:14
I get it
<bljenkins>Project fcrepo-fixity build #234: STILL UNSTABLE in 4 min 7 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-fixity/234/
<barmintor>we need to extract an interface there, and have the ISPN cluster use it's own impl for the cacheEntry16:15
this other stuff is to support non-ISPN checks
* barmintor gets back in his hole
<cbeer>barmintor: yes, we should. i was just too lazy
<barmintor>lazy+-16:16
<cbeer>yup.
<bljenkins>Project fcrepo-kitchen-sink build #349: SUCCESS in 4 min 32 sec: http://ci.projectblacklight.org/jenkins/job/fcrepo-kitchen-sink/349/
<barmintor>I have to do it now, b/c the binarystore isn't serializable, so ISPN can't transmit the ref to the cachestore
that's cool, it suggests this approach actually works16:17
:)
Related: one of the things I liked notionally about Lily was map-reduce
it turns out ISPN does map-reduce16:18
so… huzzah!
although that might also mean this service is only going to run my little job once for each binary key...16:19
* barmintor twirls his moustache
<ajs6f>You grew a mustache?
* barmintor twirls his metaphorical moustache16:20
<ajs6f>You should grow a mustache.
<cbeer>barmintor: are you doing map/reduce, or the distributed execution stuff?
<barmintor>cbeer: the distributed execution stuff
<cbeer>ok. i believe dist. ex. runs on all nodes16:21
(or, one random node, or something else.. there's some targetting stuff, i think)
<barmintor>yeah, I call .submitEverywhere
<cbeer>the actual map/reduce does the de-dupe binary keys thing
<barmintor>^^ this is not a joke, that's actually the method name
yeah, the API makes sense in that context
so, huzzah!
<pivotal-bot_>Chris Beer added "Add pagination for child node listings in rdf responses" https://www.pivotaltracker.com/story/show/5118392716:25
* ajs6f leaves17:17
* nbanks leaves18:57
* nbanks joins19:24
* nbanks leaves19:31
* nbanks joins21:24
* nbanks leaves21:29
* awoods leaves22:19
* nbanks joins23:25
* nbanks leaves23:31

Generated by Sualtam