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

Using timezone: Eastern Standard Time
* ksclarke leaves01:00
<pivotal-bot____>Andrew Woods added "Remove extraneous rdf-schema#label from "export" triples" https://www.pivotaltracker.com/story/show/6517364001:59
Andrew Woods started "Remove extraneous rdf-schema#label from "export" triples" https://www.pivotaltracker.com/story/show/65173640
* 65MAAJMKW joins02:11
[fcrepo-jms-indexer-pluggable] awoods opened pull request #29: Ensure child resources are removed from index on deletes. (master...full-clean) http://git.io/s5bBGA
* 65MAAJMKW leaves
<pivotal-bot____>Andrew Woods added comment: "Resolved with: https://github.com/futures/fcrepo-jms-indexer-pluggable/pull/29
Note: There will still rema..." https://www.pivotaltracker.com/story/show/65151810
* github-ff joins
[fcrepo-jms-indexer-pluggable] awoods created full-clean (+1 new commit): http://git.io/SeQHfw
fcrepo-jms-indexer-pluggable/full-clean 294c4fe Andrew Woods: Ensure child resources are removed from index on deletes....
* github-ff leaves
<pivotal-bot____>Andrew Woods finished "fcr:content and fcr:export triples not deleted from RDF index when parent object is deleted" https://www.pivotaltracker.com/story/show/65151810
Andrew Woods added comment: "@ajs6f, please review." https://www.pivotaltracker.com/story/show/65151810
* kaarefc joins02:26
* kaarefc leaves02:46
* fasseg joins08:53
* gregjansen joins09:22
* mikeAtUVa joins10:03
* ksclarke joins10:12
* github-ff joins10:14
[fcrepo-jms-indexer-pluggable] awoods created pom-version-update (+1 new commit): http://git.io/2EQDzA
fcrepo-jms-indexer-pluggable/pom-version-update 4d85522 Andrew Woods: Update pom project.parent version to 4.0.0-alpha-4-SNAPSHOT
* github-ff leaves
* github-ff joins10:21
[fcrepo-jms-indexer-pluggable] awoods opened pull request #30: Update pom project.parent version to 4.0.0-alpha-4-SNAPSHOT (master...pom-version-update) http://git.io/XcNTzA
* github-ff leaves
<pivotal-bot____>Andrew Woods edited "Enable index rebuild" https://www.pivotaltracker.com/story/show/6516857410:32
* mikeAtUVa leaves10:59
<awoods>fasseg?11:01
<fasseg>oops sry...omw
* scossu joins11:03
* ajs6f joins11:08
<awoods>https://wiki.duraspace.org/display/FF/Beta+Development11:12
<pivotal-bot____>Andrew Woods added comment: "https://github.com/futures/fcrepo4/pull/226" https://www.pivotaltracker.com/story/show/6108785611:18
A. "Torgeliant" Soroka edited "Fixity service does not work on Federated content" https://www.pivotaltracker.com/story/show/64112028
* github-ff joins
[fcrepo4] awoods opened pull request #226: Add FedoraFileSystemConnector, so that federated properties and fixity w... (master...federation-fixity) http://git.io/uf0sJQ
* github-ff leaves
<pivotal-bot____>A. "Torgeliant" Soroka edited "Add properties to federated nodes" https://www.pivotaltracker.com/story/show/61087856
Andrew Woods added comment: "https://github.com/futures/fcrepo4/pull/226" https://www.pivotaltracker.com/story/show/64112028
A. "Torgeliant" Soroka edited "fcr:content and fcr:export triples not deleted from RDF index when parent object is deleted" https://www.pivotaltracker.com/story/show/6515181011:21
Andrew Woods edited "Remove extraneous rdf-schema#label from "export" triples" https://www.pivotaltracker.com/story/show/6517364011:28
A. "Torgeliant" Soroka added "Inject triples for supertypes into RDF" https://www.pivotaltracker.com/story/show/6521027211:56
A. "Torgeliant" Soroka edited "Inject triples for supertypes into RDF" https://www.pivotaltracker.com/story/show/6521027211:57
* scossu leaves
* scossu joins
<pivotal-bot____>A. "Torgeliant" Soroka added "Provide pluggable strategies for updating resource properties" https://www.pivotaltracker.com/story/show/6521084212:02
<ajs6f>awoods:12:21
http://www.day.com/specs/jcr/2.0/3_Repository_Model.html#3.7.6.7 Node Type Attribute Subtyping Rules
We were completely wrong.
Primary types _can_ inherit from mixins.
So the marker-property thing is entirely unneccessary.12:22
<awoods>ajs6f: Yes. The rule is that mixins can only inherit from other mixins.12:27
<ajs6f>Right.12:28
SO scossu can make primary types that are "inherently" indexable, if that's what's watned.
<awoods>ajs6f: as long as we expose the hierarchy of types in the rdf... as planned.12:29
<ajs6f>awoods: Right. We have a ticket ready, and I don't think it should be any big deal.
https://www.pivotaltracker.com/story/show/65210272
<pivotal-bot____>feature: Inject triples for supertypes into RDF (unscheduled) / owner:
<scossu>ajs6f, awoods: that would add a great deal of power to the RDF index.12:36
* github-ff joins12:37
[fcrepo4] awoods created rdf-label (+1 new commit): http://git.io/oWmQpQ
fcrepo4/rdf-label 7b9fa9e Andrew Woods: Remove extraneous rdf-schema#label triple....
* github-ff leaves
<pivotal-bot____>Andrew Woods added comment: "Resolved with: https://github.com/futures/fcrepo4/pull/227" https://www.pivotaltracker.com/story/show/65173640
Andrew Woods finished "Remove extraneous rdf-schema#label from "export" triples" https://www.pivotaltracker.com/story/show/6517364012:38
* github-ff joins12:40
[fcrepo-jms-indexer-pluggable] awoods closed pull request #30: Update pom project.parent version to 4.0.0-alpha-4-SNAPSHOT (master...pom-version-update) http://git.io/XcNTzA
* github-ff leaves
* travis-ci joins12:55
[travis-ci] futures/fcrepo-jms-indexer-pluggable#78 (master - 4d85522 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo-jms-indexer-pluggable/compare/a9269e199aa4...4d85522f8d8a
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo-jms-indexer-pluggable/builds/18287067
* travis-ci leaves
<cbeer>awoods: on https://github.com/futures/fcrepo4/pull/227, are you saying you looked and what I linked to was not using them? or that you looked, and what I linked to is what you were looking for?13:16
<awoods>cbeer: I looked in the UI and did not see any indication of the labels... but I know helpers.getObjectTitle() uses labels, so I am giving it another look.13:17
<ajs6f>awoods: ping13:28
<awoods>ajs6f
<ajs6f>Can you rebase indexer pr 29 over master? (Master post-your merge of 30, the version update.) I'm getting:13:29
org.fcrepo:fcrepo:pom:4.0.0-alpha-3-SNAPSHOT
Or rather...
on-resolvable parent POM: Could not find artifact org.fcrepo:fcrepo:pom:4.0.0-alpha-3-SNAPSHOT
And I think it's because of the version update
<awoods>ajs6f: sure
<ajs6f>Rock on, you rockin' rocker.
<awoods>cbeer/ajs6f: The export label is indeed used in the "Export As..." drop-down.13:31
<ajs6f>cbeer/awoods: Then, if we want to get right with the RDF deities, we should introduce proper resources to represent serializations.13:32
<awoods>with the label in-place, the dropdown shows: "jcr/xml"
without the label, the dropdown shows: "http://localhost:8080/rest/test/test-image/fcr:export?format=jcr/xml"13:33
<ajs6f>The labels clearly make for a better UI (labels tend to do that) but it would be nice to offer correct RDF.13:34
In fact, there are other sorts of assertions to make about formats.
Version
Authority
Mime-type
Backwards compat or lack thereof.
Expected audience.
<awoods>ajs6f: agreed13:35
<ajs6f>"Et cetera, et cetera, et cetera…" — Yul Brynner
http://upload.wikimedia.org/wikipedia/commons/7/77/S.Kragujevic%2C_Yul_Brynner_in_Sarajevo%2C1969.JPG
Yul Brynner, looking oddly pimptastic.13:36
<awoods>ajs6f/cbeer: It sounds like the resolution here is to create a new ticket for describing serialization formats... and likely ensuring that it ties into the UI.13:38
ajs6f/cbeer: The issue initially came up because of: https://www.pivotaltracker.com/story/show/6515181013:39
<pivotal-bot____>bug: fcr:content and fcr:export triples not deleted from RDF index when parent object is deleted (finished) / owner: Andrew Woods
<ajs6f>awoods: Meaning for presenting published resources to reprsent formats?
<awoods>ajs6f: Is that a sentence?13:40
<ajs6f>No, it's a freaking triple.
"Describing serialization formats" could mean many things.
Wiki pages.
Human-readable webpages from the app.
Etc.
Do you mean, "RDF that describes the formats, presented from the app, available as sets of triples at locations each URL of which represents a format and is used to represent that format in RDF"?13:41
* github-ff joins13:42
[fcrepo-jms-indexer-pluggable] awoods force-pushed full-clean from 294c4fe to 5eb0a19: http://git.io/iGqV3Q
fcrepo-jms-indexer-pluggable/full-clean 5eb0a19 Andrew Woods: Ensure child resources are removed from index on deletes....
* github-ff leaves
<ajs6f>awoods: Thnx ^^^
<awoods>ajs6f: I rebased on master a forced a push. So you will have to pull down a new branch.
<ajs6f>awoods: Got it. git fetch should get it, right?
<awoods>ajs6f: maybe13:43
<ajs6f>Let's find out.
<awoods>fun
ajs6f: RE: "No, it's a freaking triple...", yes.13:44
Your description is what I meant.
<ajs6f>awoods: Then you should file a ticket.
That's how we get work done around here, y'now.13:45
<awoods>ajs6f: The issue in the meantime is, "Do we clean-up the Export As... dropdown with the current commit or with the new ticket"?13:46
ajs6f: Here is the line that currently uses the label: https://github.com/futures/fcrepo4/blob/master/fcrepo-http-commons/src/main/java/org/fcrepo/http/commons/responses/ViewHelpers.java#L13613:48
<ajs6f>HOw about a crappy string-manip that pulls out the value of format=, thoroughly commented as "TODO: stop doing this".
That's the Fedora wyta?
way?
<awoods>ajs6f: Likely having that logic inside of this "if" block: https://github.com/futures/fcrepo4/blob/master/fcrepo-http-commons/src/main/java/org/fcrepo/http/commons/responses/ViewHelpers.java#L14713:49
* fcrepo-bot joins13:51
<ajs6f>We do stuff like this all the time.
As long as it doesn't live longer than a few weeks… who cares?
awoods: For future ref: deleting an origin branch, then doing git fetch, then checkin g out the origin branch… that gets you the latest.13:54
I had to dance over some detached heads, but I just threw them in the trash.
That sounded weird.
<awoods>ajs6f: which is not all that uncommon coming from the ajs6f context.13:55
<ajs6f>awoods is a fountain of gentle support.13:56
* gregjansen leaves14:02
* gregjansen joins14:03
<pivotal-bot____>Andrew Woods added "RDF-ify serialization formats" https://www.pivotaltracker.com/story/show/6522140414:08
<ajs6f>awoods: Put that in the epic, "RDF all the things"14:09
<awoods>ajs6f: I am not seeing that epic14:10
* gregjansen leaves
<ajs6f>That's because you're already inside it.14:11
afk bbs
* ajs6f leaves
* ajs6f joins14:16
<pivotal-bot____>Andrew Woods added "Cache Federated SHA-1 values" https://www.pivotaltracker.com/story/show/6522232014:17
<ajs6f>awoods: about ^^^14:18
I get that we don't want to calc hashs all the time.
But do we need a way to let APi users say, "No, I really want you to go to persistence and _calc_ the hash, not just give me a value that was accurate at some time in the past."
I'm thinking about TRAC and other certifications.14:19
Is MODE federation contract even wide neough for both?
<awoods>ajs6f: the fixity service provides for that use case, no?
<ajs6f>awoods: I don't know: does it? DOes it always go to persistence? If so, cool.14:20
<awoods>The fixity service goes to disk and compares the calculated value with the stored value.
<ajs6f>So there's always a fresh calc involved when you make a fixity operation?
<awoods>yes
<ajs6f>I'm good, then.14:21
<awoods>ajs6f: It becomes a little trickier when your file gets corrupted and you want to reset the cached value.
<ajs6f>Should that operation be in the fixity service?14:22
And what about the "self-healing" shenanigans eddies introduced?
<awoods>ajs6f: I think a person needs to be involved.
self-healing deals with multiple datastores
<ajs6f>Sure, the op should emit from a person, but should it go _through_ fixity, or normal reporting?
<awoods>Here we are talking about only one copy.
<ajs6f>I thought self-healing dalt with clustering, too?14:23
dealt
<awoods>I doubt it
<ajs6f>We don't really do clustering very well right now, do we.
<awoods>clustering a fixity is another beast.
nope, there are issues.
you can update the cached property with a PATCH request to the /jcr:content node.14:24
<ajs6f>Maybe clustered fixity is as different from single-node fixity as a clustered resource is from a single-node resource.
Yeah, but that implies you _know_ what the right hash is.
<awoods>yes
<ajs6f>I'm asking about when the metadata is corrupt (or you think it is) but you think the data is correct.14:25
You retrieve a new value and manually trigger a change in the cached value?
<awoods>As it stands now, you have all of the information that you could need: you have the old cached hash, and you have the corrupted file with its different hash. The user decides.
or vice versa
<ajs6f>I know the user _decides_. I'm saying we should make it easy to implement the decision.14:26
One operation, that means: "Go get a fresh fixity and assume it is correct (aka cache it)."
Not an urgent priority, but I'll bet you a barmintor-curated beer that someone will ask for it.14:27
<awoods>Sure. My take is that we currently provide the information and given the frequency of corruption, sinking more time into making the scenario more user-friendly will not be worth the squeeze.
<ajs6f>Okay. After several major failures from our local "archival" persistence (a purchased tape robot system) I'm not sure that's true, but let's wait until we find out otherwise.14:31
awoods: what is the full-clean branch of the indexer? Should I care?14:34
<awoods>ajs6f: That is the branch you are reviewing14:35
<ajs6f>k
<awoods>...from whence pr-29 came: https://github.com/futures/fcrepo-jms-indexer-pluggable/pull/29/files14:36
<ajs6f>k— the name threw me. Sounds like I don't need to care.14:41
awoods: I'm going to have to leave the call tomorrow at 11:30: hiring committee stuff. So let's front load the call with whatever you might want me for, even if it's not a natural order?14:44
<awoods>ajs6f: sure14:45
<ajs6f>thx
* gregjansen joins14:46
* fcrepo-bot leaves14:55
* edInCo joins14:57
<ajs6f>awoods: Sorry to be a stinker about the indexer PR, but I really want to get into the mode of accreting planned work at the very granular level.15:07
* gregjansen leaves
<ajs6f>awoods: and making more comments that explain what we thought we were doing.
<awoods>ajs6f: It is good that you are the stinker in the box. I like the principle.15:08
<ajs6f>I AM THE STINKER-IN-THE-BOX!15:09
I need a good avatar for that.
That sounds like a character from old fashioned English panto.
"Oh, no, kids! Puss-in-Boots has come across a Stinker-in-the-Box! What will happen now?"15:10
<awoods>mercy
* gregjansen joins15:11
<ajs6f>Yes, Puss-in-Boots usually does better by being merciful. That tends to acquire allies, much like a good D&D session.
That's how we got to be friends with the MODE team.15:12
awoods: If you write some comments (and maybe rename those variables) I will merge straightaway.
<awoods>first lunch, then updates to rdf#label commit, then comments/variables.15:13
afk15:14
<ajs6f>Okay. I've got the rdfs:label thing in front of me, but I've also got a long afternoon meeting in front of me.
afk
* ajs6f leaves
* gregjansen leaves15:36
* github-ff joins16:51
[fcrepo4] awoods pushed 1 new commit to rdf-label: http://git.io/qYqs8A
fcrepo4/rdf-label e55a6c5 Andrew Woods: Ensure "Export As..." dropdown label still works, and add test.
* github-ff leaves
* travis-ci joins17:11
[travis-ci] futures/fcrepo4#1510 (rdf-label - e55a6c5 : Andrew Woods): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/7b9fa9e248ab...e55a6c50ae9e
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/18303799
* travis-ci leaves
<pivotal-bot____>Andrew Woods edited "RDF-ify serialization formats" https://www.pivotaltracker.com/story/show/6522140417:39
* ksclarke leaves18:35
* edInCo leaves19:36
* scossu leaves19:40
* ksclarke joins20:24
* ksclarke leaves
* ksclarke joins20:27