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

Using timezone: Eastern Standard Time
* ksclarke1 leaves01:46
<pivotal-bot_____>Frank Asseg added "Add platform profile for cluster Timisoara University" https://www.pivotaltracker.com/story/show/6745606804:08
Frank Asseg started "Add platform profile for cluster Timisoara University" https://www.pivotaltracker.com/story/show/67456068
* mikeAtUVa joins09:04
* gregjansen joins09:11
* gregjansen leaves09:26
* ermadmix joins09:27
* scossu joins09:33
* ajs6f1 joins10:04
* ajs6f leaves10:08
* ajs6f1 leaves10:09
* ajs6f joins10:10
* awoods joins10:13
* scossu leaves10:17
* scossu joins
<ajs6f>cbeer: Do we have the "use a stored transform" functionality of fcrepo-transform documented anywhere? I can't find it...
I mean I can't find documentation.10:18
<awoods>ajs6f: Does this help? https://wiki.duraspace.org/display/FF/RESTful+HTTP+API+-+Transform10:19
<ajs6f>awoods: No, that's my point. That just shows the use of a default transform or a request-supplied transform, not a stored transform.10:20
<awoods>ajs6f: I do not recall the existence of such documentation10:21
<ajs6f>awoods: Isn't that what Ronald Reagan would say?10:22
http://articles.chicagotribune.com/1990-02-23/news/9001160085_1_john-poindexter-recall-reagan-testimony
* escowles joins
* ajs6f1 joins10:25
* ajs6f leaves
* ksclarke joins10:26
* ajs6f1 leaves10:37
<pivotal-bot_____>Esme Cowles started "Use Case: Ingesting large files into the repository" https://www.pivotaltracker.com/story/show/6715451010:40
* ajs6f joins10:42
<pivotal-bot_____>Andrew Woods added comment: "I suspect that getting a datastream onto localhost (or some network filesystem) would satisfy this use case..." https://www.pivotaltracker.com/story/show/6715451010:43
Eric James added comment: "One last stab at recreating the persisting projections, and I was able to recreate.. However when I removed ..." https://www.pivotaltracker.com/story/show/6730811010:57
Esme Cowles added comment: "I think this is basically ready for validation. I've added a couple of notes to the wiki about setting java..." https://www.pivotaltracker.com/story/show/67154510
Andrew Woods added comment: "Maybe this ticket? #65600152 , @escowles " https://www.pivotaltracker.com/story/show/6715451010:59
Esme Cowles added comment: "I think that would be a good IT to have, but it's also probably good to actually test with 1TB files to make..." https://www.pivotaltracker.com/story/show/6715451011:00
* edInCo joins
awoods: Can't wait to see the summaries!11:01
* awoods leaves11:03
<escowles>ajs6f: committers' call?11:05
* awoods joins
* ajs6f leaves11:06
* ajs6f joins11:07
* nbanks joins11:08
* martind_ joins11:09
<nbanks>https://wiki.duraspace.org/display/FF/Design+-+New+Model+Validation+Workflow11:18
* gregjansen joins
* gregjansen leaves11:21
<awoods>ajs6f: https://www.pivotaltracker.com/story/show/6715537011:23
<pivotal-bot_____>feature: Use Case: Structural Validation (unstarted) / owner:
<edInCo>Does the proposal not take away functionality, only add to what is already there?11:32
<ajs6f>edInCo: To my understanding, yes.11:33
<edInCo>ajs6f: OK, that helps for the notes!11:34
<ajs6f>edInCo: NP!
scossu: I'm going to wait to hear your thoughts— you'll put them in comments on that wiki page?11:41
Or on the page with your use case?
<pivotal-bot_____>A. "Ropsamingo" Soroka added "Make Transform and TransformFactory in fcrepo-transform use generics instead of reflection" https://www.pivotaltracker.com/story/show/6748387811:43
A. "Ropsamingo" Soroka started "Make Transform and TransformFactory in fcrepo-transform use generics instead of reflection" https://www.pivotaltracker.com/story/show/67483878
<scossu>ajs6f: added attachment in https://wiki.duraspace.org/display/FF/Art+Institute+of+Chicago+Use+Case+-+Structural+Validation11:46
<ajs6f>scossu: thnx!
<scossu>ajs6f: Will also post my thoughts on your proposal page.
<ajs6f>scossu: Excellent. I'll reply to them there. We can make a new page as a child of your use case page, if we feel that is worthwhile.11:47
<scossu>ajs6f: ok.11:48
<nbanks>https://github.com/futures/ontology12:01
gtg Another meeting
<edInCo>What is the status of 3.7.2 ?12:03
any unofficial activities at c4l 2014?
<awoods>edInCo: barmintor indicated that he would be working on getting 3.7.2 around c4l.12:05
<edInCo>awoods: Oh yeah, I remember him saying that12:07
<awoods>thanks for taking notes today, edInCo.12:10
<pivotal-bot_____>Eric James added "changing an existing externalSource name fails Context initialization" https://www.pivotaltracker.com/story/show/6748690812:18
Eric James added comment: "ticket for the above comment: https://www.pivotaltracker.com/s/projects/684825/stories/67486908" https://www.pivotaltracker.com/story/show/67308110
<ajs6f>scossu: The domain and ranges of properties on that diagram don't seem to be indicated anywhere, and it's not clear to me what the arrows mean. Are they subtyping/inheritance?12:52
scossu: Sorry, obviously the domains are clear, it was the ranges I meant to say are not indicated.12:53
<pivotal-bot_____>A. "Ropsamingo" Soroka added "Attempt to instantiate an AIC use case in the "new model validation"" https://www.pivotaltracker.com/story/show/6749034412:54
A. "Ropsamingo" Soroka started "Attempt to instantiate an AIC use case in the "new model validation"" https://www.pivotaltracker.com/story/show/67490344
<scossu>ajs6f: I didn't show the domain and range in that drawing for display purposes. I can attach another drawing with RDF relationships, but it's older and it doesn't match that document, so I'll have to update it.12:56
<ajs6f>scossu: If you think the differences are important okay, but after all, we're just trying to do an example here, so please don't go to any great trouble.
<scossu>ajs6f: there is a legend in the lower left, although "subclass" obviously meant "sub-type"12:57
<ajs6f>scossu: So there is! Somehow I missed it...
* ermadmix leaves13:14
<ajs6f>scossu: I'm having a lot of trouble understanding why so few of these types are mixins, and why so few of the mixins have any properties associated to them. It's not at all clear to me why it is so important for a resource to possess only certain types at once (i.e. for so many of the types to be formally disjoint). Is this an effect of some application layer functionality leaking into the repository?13:25
* nbanks leaves13:28
* gregjansen joins13:32
* nbanks joins13:35
* ermadmix joins13:49
* ksclarke leaves14:15
* ksclarke joins14:18
* nbanks leaves14:52
<scossu>ajs6f: an exclusive type model works for the collection management system we have and the diverse departments that use it; my choice was to use primary types to define resources in a OO style, and mixins to define transversal properties such as publishing settings.14:56
<ajs6f>scossu: Okay, so the character of your model is coming from a desire to mirror your organizational structure?14:57
<scossu>ajs6f: but regardless of how heavy the use of primary types, my point is that they looked very useful not only for my use case, but for many others.
<ajs6f>scossu: I disagree— _restrictive_ types, sure. They wold be useful. I don't think primary types are useful in and of themselves.
It does not follow from a (legit) desire for restrictive types that primary types are the thing to use.14:58
<scossu>ajs6f: I have to say that I modeled that structure following the JCR specs because at that time it looked like something that Fedora would support (as of alpha 1).14:59
<ajs6f>scossu: In _no_ way am I suggesting that you have done anything wrong here.
scossu: Quite the contrary. It's your work that has helped us get this far.
scossu: The questions I'm raising are about where we want to go from here, not at all about the work you've done to get here.15:00
scossu: We've got another factor at play. You are asking for disjoint types as well as restrictive types. Both are completely reasonable desires for which I'm sure there will be plenty of use cases (although not from folks migrating from F3, which never offered disjoint typing and the restrictive typing scheme of which was fairly primitive). We nay need to think about different facilities for each, but it is (in my mind) completely sett15:02
<scossu>ajs6f: If I can achieve the same results in another way, I'm fine! e.g. we can also define a rdf:type fedora:exclusive, and base our validation on that.15:03
e.g. aic:image rdf:type fedora:exclusive
<ajs6f>No, we wouldn't do that. We would connect the type to the meta-quality, e.g.
Oh, wait, is aic:image a resource or a type?15:04
<scossu>type
<ajs6f>Oh, okay. Then yeah, we could assign it at that level.
<scossu>Yeah, we are talking about the same thing: building a super-structure for node types.15:05
<ajs6f>Although it might be easier to simply have all restrictive types subtype a "fedora:exclusive" type, which would prevent the kind of meta-typing you're doing in the above triple.
Well, sort of. You're talking about annotating a type.
I'm talking about adding a type to the same type system, just one with a special meaning.
<scossu>that's right.15:06
<ajs6f>You want to do:
http://en.wikipedia.org/wiki/Kind_(type_theory)
but I don't want to because the code tends to be a little funkier.
Anyway, we're trying to get to the same place.
<scossu>I'm glad to hear that.
<ajs6f>Call me when you get there so you can give me directions.15:07
Actually, I wonder if "is disjoint with other types except its own supertypes and subtypes and mixins" is a monad? Oh, well, doesn15:08
't really matter, I guess.
<scossu>I'll reply to your comment on the wiki, for the public records, and I'll give you a call in a little bit.15:09
<ajs6f>scossu: Let me go peer at the code a bit, and see whether we can introduce this "speciality" and where and how much it might cost.
A call?
* nbanks joins
<scossu>I thought that's what you meant.
<ajs6f>Oh, no! Sorry, scossu, that wa a joke.
<scossu>ah! I'm really bad at getting jokes15:10
* nbanks leaves
<ajs6f>IOW: "I don't know where I'm going, so when you get to where we're both trying to go, tell me how to get there."
It's hard to hear jokes in IRC.
<scossu>ajs6f: Now I get it. I was trying to get past the 2 previous days of sickness and entertain the baby while I made my RDF reasoning.15:11
<ajs6f>scossu: We have validated the joke, and we annotated with "?j toldBy ajs6f. ?j a bad:joke"15:16
scossu: So the things I think we can say are demanded by your use cases are:15:17
1) Disjoint types.
2) Property-restricting types.
3) Child-node-restricting types.
and I think that's it...15:18
Oh, wait, no.
You also want:
4) Property-mandating types.
and
5) Child-node-mandating types.
The SPARQL idea will cover the last 4 without difficulty. Disjointness is giving problems because disjointness is not a quality of classes that can be expressed without OWL or something similarly powerful.15:20
So we come to our trick of having a special property or superclass.
Now I will go look and see where we would need to impl that. Tell me if you think the above summary is in any way wrong or insufficient.
<scossu>ajs6f: that looks like it's gonna cover a lot of use cases.15:21
<ajs6f>That's what you pay us for.15:22
<scossu>ajs6f: and I assume that I can create RDFS sub-types, right?15:23
ajs6f: it's worth every cent.15:24
<ajs6f>scossu: I would ideally like to be using RDFS as the language. cbeer already did most of the work for that.
It exists in the API now. There's no reason not to use it, I don't think.15:25
<scossu>ajs6f: so I can still start with a very restrictive top-level type and define properties and child nodes as I go down the hierarchy?15:26
(you tell me if I'm getting fixated with this hierarchy thing)
<ajs6f>scossu: I want to separate the notions of restriction and mandatoriness. In other words, from any different point in the type system, one (ideally) would be able to add a type and either restrict more properties/child-nodes or mandate more of them. If you like to organize that into a hierarchy and use the disjointness facility (whatever it ends up being), fine by me. Other people will want to use a "flatter" approach with lots of mixins15:28
scossu: This is one reason I wanted to use SPARQL (or something like it). It is easy to express either mandatoriness or forbiddeness in that way.15:30
JCR's type system is biased towards "allowance".
Monotonic increase in permissiveness.
* ksclarke leaves15:33
<ajs6f>afk bbs
* ajs6f leaves15:37
* ajs6f joins15:38
scossu: Added a note on:15:46
https://wiki.duraspace.org/display/FF/Design+-+New+Model+Validation+Workflow
summarizing our thinking about disjointness.
<scossu>ajs6f: me too!15:47
* scossu1 joins16:34
* scossu leaves
<awoods>escowles_lunch: still eating?16:35
* github-ff joins16:38
[fcrepo4] ajs6f created GenericFcrepoTransform (+4 new commits): http://git.io/a6iyMw
fcrepo4/GenericFcrepoTransform 57032ce ajs6f: Simplifying type of LDPath transform result
fcrepo4/GenericFcrepoTransform 2f5b67c ajs6f: Cleaning up exceptions
fcrepo4/GenericFcrepoTransform d23f4e4 ajs6f: Eliminated reflection
* github-ff leaves
* github-ff joins16:39
[fcrepo4] ajs6f opened pull request #266: Generic fcrepo transform (master...GenericFcrepoTransform) http://git.io/fShyIw
* github-ff leaves
<pivotal-bot_____>A. "Ropsamingo" Soroka added comment: "https://github.com/futures/fcrepo4/pull/266" https://www.pivotaltracker.com/story/show/67483878
A. "Ropsamingo" Soroka finished "Make Transform and TransformFactory in fcrepo-transform use generics instead of reflection" https://www.pivotaltracker.com/story/show/67483878
<ajs6f>awoods: ^^^ Just a simple change to fcrepo-transform to eliminate the use of reflection for a performance improvement
afk16:40
* ajs6f leaves
* travis-ci joins17:00
[travis-ci] futures/fcrepo4#1623 (GenericFcrepoTransform - 7f5a343 : ajs6f): The build passed.
[travis-ci] Change view : https://github.com/futures/fcrepo4/compare/57032ce7e836^...7f5a343f12c6
[travis-ci] Build details : http://travis-ci.org/futures/fcrepo4/builds/20719126
* travis-ci leaves
* mikeAtUVa leaves17:01
* gregjansen leaves17:38
<escowles>awoods: forgot to change it when i got back... but then went to pickup the kids, so kind of...17:46
<awoods>escowles: are you available now-ish, or is tomorrow better for a chat?17:47
<escowles>i can talk for a few minutes
<awoods>escowles: see you on the standard hangout?17:48
<escowles>yep, i'm firing up chrome now
* ermadmix leaves17:57
* edInCo leaves18:48
* scossu1 leaves19:27
* scossu joins19:58
* scossu leaves20:01
* scossu joins20:30
* pivotal-bot_____ leaves20:54
* pivotal-bot_____ joins20:56
* pivotal-bot_____ leaves21:06
* pivotal-bot_____ joins
* pivotal-bot_____ leaves21:15
* scossu leaves21:19
* pivotal-bot_____ joins
* pivotal-bot_____ leaves21:23
* pivotal-bot_____ joins21:32
* pivotal-bot_____ leaves21:39
* pivotal-bot_____ joins
* pivotal-bot_____ leaves21:43
* pivotal-bot_____ joins21:44
* pivotal-bot_____ leaves22:08
* pivotal-bot_____ joins
* martind_ leaves23:36

Generated by Sualtam