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

Using timezone: Eastern Standard Time
* ksclarke leaves01:32
* edInCo leaves01:40
* github-ff joins01:43
[fcrepo4] cbeer pushed 1 new commit to master: http://git.io/w3PLcg
fcrepo4/master f9edd2f Chris Beer: support the Atom Slug: header to provide the name of the node to create on POST to an existing collection node
* github-ff leaves
* travis-ci joins01:58
[travis-ci] futures/fcrepo4#940 (master - f9edd2f : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/88bfddc8c7a4...f9edd2f47e82
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/9634508
* travis-ci leaves
<bljenkins>Project fcrepo-fixity-corrupter build #211: SUCCESS in 3 min 18 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/211/02:26
Project fcrepo-kitchen-sink build #478: STILL UNSTABLE in 10 min: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/478/02:36
* fasseg joins05:10
* github-ff joins05:18
[fcrepo4] fasseg tagged fcrepo-4.0.0-scape at c88c63b: http://git.io/iSQnug
* github-ff leaves
<fasseg>I added this tag so i had a stable version as a backend for the SCAPE project05:23
in case you were wondering...
<bljenkins>Project fcrepo-fixity-corrupter build #212: SUCCESS in 2 min 47 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/212/06:04
Project fcrepo-kitchen-sink build #479: STILL UNSTABLE in 7 min 18 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/479/06:11
* fasseg leaves07:25
* fasseg joins
* fasseg leaves
* ksclarke joins08:20
* edInCo joins08:43
* edInCo1 joins09:30
* edInCo leaves09:33
<pivotal-bot>Andrew Woods edited "make islandora work with latest build of fcrepo4. " https://www.pivotaltracker.com/story/show/4835140109:36
Andrew Woods edited "make islandora work with latest build of fcrepo4. " https://www.pivotaltracker.com/story/show/48351401
Andrew Woods edited "Add transactions support to Tuque" https://www.pivotaltracker.com/story/show/4901079709:37
Andrew Woods edited "Add transactions support to Tuque" https://www.pivotaltracker.com/story/show/49010797
Andrew Woods added "Create and Install SSL Cert: Sonar, Nexus, Jenkins" https://www.pivotaltracker.com/story/show/5428381210:22
Andrew Woods edited "Upgrade to Jersey 2.0" https://www.pivotaltracker.com/story/show/5220082310:24
Andrew Woods edited "Upgrade to Jersey 2.0" https://www.pivotaltracker.com/story/show/52200823
Andrew Woods edited "Create and Install SSL Cert: Sonar, Nexus, Jenkins" https://www.pivotaltracker.com/story/show/5428381210:26
Andrew Woods added "Migrate from Gluck to AWS" https://www.pivotaltracker.com/story/show/54284208
Andrew Woods edited "Replace Velocity templating with LDPath's Freemarker integration" https://www.pivotaltracker.com/story/show/5306415510:29
Andrew Woods added "Fair Game" https://www.pivotaltracker.com/story/show/5428520210:40
Andrew Woods edited "Tomcat 7 on futures6 Permgen Exception on re-deploy" https://www.pivotaltracker.com/story/show/4654330110:42
* jongibson joins10:44
<pivotal-bot>Andrew Woods added "Postponed" https://www.pivotaltracker.com/story/show/5428558610:45
Andrew Woods edited "Published events are missing a resolvable location for the resources" https://www.pivotaltracker.com/story/show/5286074910:46
Andrew Woods edited "Index projected nodes into MODE's Lucene index" https://www.pivotaltracker.com/story/show/5074867710:47
Andrew Woods edited "Create a sequencer that copies projected bagit nodes to internal nodes." https://www.pivotaltracker.com/story/show/50748291
Andrew Woods edited "Implement Write Operations in BagIt Federation Connector" https://www.pivotaltracker.com/story/show/4876429510:48
Andrew Woods edited "Fixity service should send a JMS message when fixity runs on an object" https://www.pivotaltracker.com/story/show/46461909
* gregjansen1 joins11:00
* gregjansen leaves11:02
* gregjansen1 leaves11:04
<pivotal-bot>Yuqing Jiang started "make islandora work with latest build of fcrepo4. " https://www.pivotaltracker.com/story/show/4835140111:07
Yuqing Jiang started "Add transactions support to Tuque" https://www.pivotaltracker.com/story/show/49010797
* edInCo1 leaves11:41
<awoods>osmandin: I assume you have looked at the module: fcrepo-storage-policy?11:48
The idea came up today regarding configuration info that does/will reside within the repository. It seems like a consistent jcr location would be in order.11:52
The only pattern I currently see for this type of content is under the fcrepo-transform project, where "/fedora:system/fedora:transform" is used.11:54
Does "/fedora:system/<scope>" seem like a reasonable pathway?
* github-ff joins11:55
[fcrepo4] cbeer pushed 4 new commits to master: http://git.io/ARKe4A
fcrepo4/master 640ccac Chris Beer: remove the /objects node from our test suite.
fcrepo4/master aa99c18 Chris Beer: use PATCH method to send SPARQL-Update queries to the resource
fcrepo4/master 1758b87 Chris Beer: remove POST with a SPARQL-Update payload (implementations should use PATCH instead), and support creating objects with SPARQL-Update or RDF payloads
* github-ff leaves
<pivotal-bot>Andrew Woods unstarted "Add transactions support to Tuque" https://www.pivotaltracker.com/story/show/4901079711:56
Chris Beer added "OPTIONS support on LDP Resources" https://www.pivotaltracker.com/story/show/5429290412:03
Chris Beer edited "OPTIONS support on LDP Resources" https://www.pivotaltracker.com/story/show/54292904
* travis-ci joins12:15
[travis-ci] futures/fcrepo4#941 (master - 778fb56 : Chris Beer): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/f9edd2f47e82...778fb568e59f
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/9654385
* travis-ci leaves
* ksclarke leaves12:23
<osmandin>awoods: that sounds reasonable for now. will think and investigate further12:25
* edInCo joins
<bljenkins>Project fcrepo-fixity-corrupter build #213: SUCCESS in 2 min 35 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/213/12:46
Project fcrepo-kitchen-sink build #480: STILL UNSTABLE in 6 min 12 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/480/12:53
<awoods>cbeer: Do you have a moment to chat?13:01
* cbeer leaves13:06
* cbeer joins13:09
<pivotal-bot>Andrew Woods added comment: "https://docs.jboss.org/author/display/MODE/Backup+and+restore" https://www.pivotaltracker.com/story/show/4610822913:11
<awoods>bbi ~1.5 hours
* pivotal-bot leaves13:23
* pivotal-bot_ joins13:28
* pivotal-bot__ joins13:35
* pivotal-bot_ leaves13:38
* pivotal-bot__ leaves13:43
* pivotal-bot__ joins
* jongibson leaves14:33
* ajs6f joins
awoods/cbeer: ping?14:34
<cbeer>ajs6f: pong
<ajs6f>cbeer: I was going to fool around with/finish the translation of the triple generation contract (for extensibility/wire-ibilty), but I'm going to wait until you/awoods/al. have firmed up the nature of JCR-housed config.14:35
(Which, to my mind, is of the first importance.)14:36
<cbeer>ajs6f: no news from me14:40
<ajs6f>That's cool. This is something that ought to be done right the first time.
* ajs6f leaves14:51
* ajs6f joins14:55
<awoods>ajs6f: The "/fedora:system/<scope>" approach seems reasonable. Do you have an alternate suggestion?15:01
<ajs6f>For the oauth scope recordation?15:02
Or for cofig scopes?
config
<awoods>I was suggesting a general location pattern for config details.15:03
<ajs6f>Right. Yeah, that sounds right. Do we want to privilege some scopes to the core via namespace, i.e.15:04
fedora:system/fedora:someconfigscope?
Or just establish the pattern and let the good times roll?
* jongibson joins
<awoods>I would suggest it depends on the feature15:05
solidly core features could have a privileged namespace
<ajs6f>So… what would be the core features that deserve protection?
^^^
Triple generation?
<awoods>I am not sure we have come across one yet
<ajs6f>Auth config?
Okay. So we reserve the right to namespace some pieces of config out of instance-mutability?15:06
But choose to not yet exercise it?
<awoods>yes
<ajs6f>I'm cool with that.15:07
<awoods>as for auth... it is clear that some folk definitely do not want auth config in the repo.
so reserving a namespace for them seems, unfriendly.
<ajs6f>Really? I thought some folk definitely do not want auth config only available through the repo.
I didn't know anyone had a problem with it being available through the repo but also available by other means.
(Chosen by the integrator.)15:08
<awoods>When sprater gets tied into the conversation, we will hear the other perspective.
<ajs6f>I thought we heard that yesterday...?
<awoods>yes and no
<ajs6f>Are you saying Scott is schizoid?15:09
<awoods>sprater was on the line much before you joined.
<ajs6f>'Cause he's always seemed pretty stable to me.
<awoods>I think he is a rock.
<ajs6f>Scott Prater rocks as hard as anyone from Wisconsin possibly can.15:10
I don't want to put anything in the privileged category now.
I want to be able to do that when we need to.
<awoods>sounds right
<ajs6f>Okay.
<awoods>rock on
<ajs6f>So… e.g. fedora:system/triplegeneration?
<awoods>or "/fedora:system/fedora:triplegeneration"15:12
<ajs6f>So… triple generation _is_ privileged?
Or does the fedora: namespace carry no semantic?
(Other that, we wrote it, not you.)15:13
than
<awoods>I was assuming "privileged" meant that a namespace was automatically registered on repo startup.
what does "privileged" mean to you?15:14
<ajs6f>I was suggesting it as meaning that integrators/users cannot mutate it. Or its meaning. Or its inheritance. Or what comes within its remit by indirect means.
So fedora:system/fedora:triplegeneration is not available to API manipulation.15:15
(Which would be, in my mind, a bad example.)
(It should be.)
Autoregistered is a useful quality, but it's not the same as "privileged", to me.15:16
It's a matter of convenience vs. a matter of a quality that persists for the lifetime of a repo.
<awoods>Can we not allow a namespace to be "privileged" in your definition for installations that use that feature but not privileged for installations that do not?
<ajs6f>I think that's a confusing road.
It points to a deeper problem:15:17
we have no genuine modularity.
To choose a feature or choose not-a-feature implies some mechanism by which that choice is made real.
Currently, the closest thing we have is Spring XML.15:18
And lots of Maven trickery.
These two things do not a modularization framework make.
But that's another epic.
<awoods>Of course we can not foresee all potential namespaces that will be needed in the future...15:19
<ajs6f>But it's not unrelated: there's a reason that OSGi chose to make Configuration Admin a core service.
So, we can't.
No. we can't. But we can reserve some (one).
If we think that we should.
Maybe nothing about the agglomeration of services that composes a running repo is sacred?15:20
Maybe it's _all_ a matter of choice?
That's not unreasonable, to ,e.
me
<awoods>I am trying to sort out where the line of your reasoning is between philosophical and practical...15:21
<ajs6f>cbeer and barmintor can tell you that such a quest is doomed to noble failure.15:22
<awoods>I can accept that
<ajs6f>Or to put it more usefully, the line is somewhere between one beer and another.
I really think we can declare all config to be mutable.15:23
I just want us to be clear about what we mean by "fedora:".
<awoods>and that it should generally be found within the repo at: /fedora:system/fedora:<scope>15:24
how many PBs did SCAPE have in mind?15:25
<ajs6f>Really? So my new insane OWL Full inference extension module and Joe Blow's genuinely useful bitstream characterization module and the kernel itself all have just the same right use that namespace?
Because it is a namespace. it's not just a prefix for identifiers.
<awoods>it goes back to your question of what "fedora:" means...15:26
^^^ PB question
<cbeer>why does it need to be fedora:<scope>? just (some module-defined namespace):<scope> should be ok, right?15:27
<ajs6f>+1
My question is: do we want to declare now
<awoods>what about the system portion: /fedora:system/...
<ajs6f>Right.
What does "fedora:" actually mean?
Besides "the committers did this".15:28
<cbeer>DEREFERENCE ALL THE THINGS!
<ajs6f>Right.
<cbeer>oh. info:fedora/
<ajs6f>We're fixing that.
<awoods>I would suggest that in this context, "fedora:" means that it is required by core fedora.15:29
<ajs6f>Then nothing we've done qualifies, I think.15:30
<cbeer>i'd define "fedora:" as just not your data in your repository… i guess that gets blurry when we start putting transforms and services in a fedora: namespace though
maybe.. architecture-specific configuration magic?15:31
<ajs6f>Right. Are there any transforms or services that are core? The kernel, to my mind, is the definition of core.
<cbeer>core, no. default, maybe?
<ajs6f>Sure. But default is different.
Default is instantiated in build and publication.15:32
Core is instantiated in code.
You can't start a repo without core.
<awoods>I can see core needing a place for config details...15:33
<ajs6f>You have to make a choice(s) to start a repo without default.
Sure, and that's where "fedora:system/fedora:*" makes sense.
<awoods>there needs to be a place that can be relied upon to exist and not be overwritten
yes, "fedora:system" seems reasonable for such a place15:34
<ajs6f>That's "privileged" /fedora:system/fedora:*.
No, fedora:system:fedora:*, not fedora:system:*, or no?
fedora:system/* implies
this is not content, it's config.15:35
<awoods>different thread: does anyone know the PB size of the SCAPE collection?
<ajs6f>fedora:system/fedora:* implies that this is privileged config that shouldn't be altered.
What is a PB? Petabyte?
<awoods>yes
<ajs6f>Lots.
Or maybe 1.
<awoods>that covers the range15:36
<ajs6f>eddies promised us billions and billions of items.
But he never gave a figure for bytes.
afkbbl15:49
* ajs6f leaves
* ajs6f joins15:51
* gregjansen joins15:52
<pivotal-bot__>Andrew Woods edited "Update policy driven storage to move configuration from Spring into the JCR tree." https://www.pivotaltracker.com/story/show/4905979916:08
<osmandin>Trying to add to fcrepo-policy. It seems that fcrepo-kernel has dependency fcrepo-storage-policy, so it seems that nothing can be called from kernel due to cyclical dependency (unlike fcrepo-transform)16:19
<awoods>osmandin: any reason not to directly update fcrepo-storage-policy?16:21
osmandin: It would be ideal if you put you work on a branch.
your
<osmandin>kernel would be a cyclical dependency..because storage-policy is kernel's dependency16:22
yes, it'll be in my forked branch16:23
* barmintor leaves16:24
<awoods>osmandin: do you actually see where kernel uses storage-policy?
<gregjansen>kernel would probably use it to generate a storage hint that is passed to the binary store logic16:26
<osmandin>awoods: I think I saw it in datastreamservice
<jongibson>awoods: did we settle on, is our call at 4pm central or 6pm?
<awoods>yes to all. PolicyDecisionPoint is just is a few places.
jongibson: our call is at 6pm ET (5pm CT)16:27
<jongibson>awoods: thx, i'll be there
<awoods>It would be nice to break that dependency from kernel16:28
<gregjansen>awoods: CDI?
<awoods>Any injection would do.
<ajs6f> CDI +116:29
<awoods>I may want to turn PolicyDecisionPoint into an interface
an have that interface in a module other than fcrepo-storage-policy
<gregjansen>I'd also support any move to use the term "policy" less often.
<ajs6f>Is that our new pilcy?16:30
<gregjansen>the new word is "pilcy"
<awoods>how about paul-see
<ajs6f>Too easy to confuse with Paul from DGI.
<osmandin>to confirm, it seems that org.fcrepo.binary.PolicyDecisionPoint in kernel is the only reference16:31
in DataStreamService.java
<awoods>osmandin: yes, PDP is the only class I see in kernel or any other module.16:32
<ajs6f>Shouldn't SOnary be telling us about this sort of thing?16:33
Sonar
<awoods>in Datastream and DatastreamService
fyi: http://sonar.fcrepo.org/
<ajs6f>awoods: I don't see anything there. It's inspecting code, and it reports out package tangles, but not library loops.16:34
Maybe there's Maven wizardry we can employ?16:35
<awoods>osmandin is creating a new module
that introduces a cycle.16:36
<pivotal-bot__>Gregory Jansen started "Create Strawman AuthN/Z Design" https://www.pivotaltracker.com/story/show/54250636
<ajs6f>I thought he was saying that a cycle already exists?16:37
<pivotal-bot__>Gregory Jansen added comment: "I've finishing moving many of the requirements found in pivotal to the strawman and fleshed out the basic..." https://www.pivotaltracker.com/story/show/5425063616:38
<osmandin>No, it' doesn't exist but to add any code to storage policy inntroduces it
otherwise it would be compiling16:39
it wouldn't be, imean
<ajs6f>Oh… okay. So what we have is some unclearness as to the structure at the very bottom.
* ajs6f leaves16:45
* ksclarke joins16:46
* osmandin leaves
<awoods>gregjansen: which jaxrs2.0 features are you waiting on?16:57
<gregjansen>awoods and others: I'd welcome feedback on this authz draft.. https://wiki.duraspace.org/display/FF/Authentication+and+Authorization
awoods: I will see if I can avoid it. I'd like to do a global interception of our REST calls and I thought that required 2.016:58
if I can do that, then I can add the pep without any changes to http-api
<awoods>you should be able to do that with servlet filters, yes.16:59
<gregjansen>awoods: tho, there are liabilities to configuration by dropping in a jar..
servlet filters don't get all our nifty resource annotations17:01
I was hoping not to replicate much of the request processing that happens at the JAX-RS level
<awoods>sure, I was just saying that changing the api should not be required.
<gregjansen>i don't get it17:02
oh the jax-rs 2.0 api, yeah I think you're right
<awoods>oh... you were talking about fcrepo-http-api17:03
<gregjansen>yeah, trying to piggyback on our fcrepo-http-api where possible in terms of grokking the request into a specific kind of action..17:04
* daniellamb leaves17:05
<gregjansen>a javax.ws.rs.ext.ReaderInterceptor might do the trick. I'
m experimenting
<pivotal-bot__>Andrew Woods added comment: "Greg's strawstuff: https://wiki.duraspace.org/display/FF/Authentication+and+Authorization" https://www.pivotaltracker.com/story/show/5425063617:07
Andrew Woods edited "Add support for nested bagit connector data directories" https://www.pivotaltracker.com/story/show/53771753
<gregjansen>awoods: thx17:08
<pivotal-bot__>Andrew Woods edited "Add support for nested bagit connector data directories" https://www.pivotaltracker.com/story/show/53771753
Andrew Woods edited "BagIt export module should send content as application/zip" https://www.pivotaltracker.com/story/show/47213685
* gregjansen leaves17:09
<pivotal-bot__>Andrew Woods added "Add ability for bagit connector to read from zipped bags" https://www.pivotaltracker.com/story/show/54318498
Andrew Woods added "Add ability for bagit connector to write to exploded bags" https://www.pivotaltracker.com/story/show/5431867017:10
* ksclarke leaves18:56
* jongibson leaves19:13
* edInCo leaves19:21
* ksclarke joins19:33
* github-ff joins19:36
[fcrepo4] cbeer created new-node-location-fcr-content (+1 new commit): http://git.io/OU5q8w
fcrepo4/new-node-location-fcr-content 8306cfe Chris Beer: make sure we send the right Location header back when POSTing binary content to create a new node
* github-ff leaves
* github-ff joins19:39
[fcrepo4] cbeer opened pull request #105: make sure we send the right Location header back when POSTing binary content to create a new node (master...new-node-location-fcr-content) http://git.io/WGKZWA
* github-ff leaves
<bljenkins>Project fcrepo-fixity-corrupter build #214: SUCCESS in 45 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-fixity-corrupter/214/19:51
Project fcrepo-kitchen-sink build #481: STILL UNSTABLE in 2 min 25 sec: http://ci.fcrepo.org/jenkins/job/fcrepo-kitchen-sink/481/19:53
* daniellamb joins20:50
* awoods leaves20:51
* daniellamb leaves20:55
* edInCo joins21:45
* daniellamb joins22:38
* daniellamb leaves22:43