Log of the #fcrepo channel on chat.freenode.net

Using timezone: Eastern Standard Time
* dwilcox joins01:34
* dwilcox leaves02:13
* dbernstein leaves02:43
* thomz joins02:59
* dwilcox joins03:02
* dwilcox leaves03:15
* github-ff joins07:46
[fcrepo4] ajs6f closed pull request #968: FCREPO-1868: remove dependency on javax.jcr.Session in kernel-api (master...FCREPO-1868) https://git.io/vEBYM
* github-ff leaves
* mohamedar joins08:00
* ajs6f joins08:40
* youn joins08:49
* bseeger joins
* westgard joins08:51
* whikloj joins09:00
* acoburn joins09:04
ajs6f: I have a question for you about rdf entailment semantics09:05
ajs6f: take, e.g., dcterms:extent, which has a range of some dcterms class09:06
<ajs6f>acoburn: I have a nagging suspicion that the world around me is actually a "Truman Story"-style mockup. Let's see if we can meet in the middle.
<acoburn>ajs6f: their examples show uses like: <foo> dcterms:extent [ rdf:value "some literal" ]09:07
<ajs6f>acoburn: That's kind of gross, but seems liegit.
<acoburn>ajs6f: and in their docs, they say that it's intended for use with non-literal values
<ajs6f>acoburn: Only non-literals can have rdf:types.
<acoburn>ajs6f: yes, but given simple RDF entailment, isn't it also the case that <a> :p "abc"^^<someClass> === <a> :p _:b0. _:b0 rdf:type <someClass>
<ajs6f>"The https://www.w3.org/TR/2004/REC-rdf-schema-20040210/#ch_domain of rdf:type is https://www.w3.org/TR/2004/REC-rdf-schema-20040210/#ch_resource. The https://www.w3.org/TR/2004/REC-rdf-schema-20040210/#ch_range of rdf:type is https://www.w3.org/TR/2004/REC-rdf-schema-20040210/#ch_class."
acoburn: You'll have to rewrite that. It came across as a bunch of smiley faces. (Maybe you are using parentheses?)09:09
<acoburn>ajs6f: let me try again
<ajs6f>acoburn: Wait, what is "abc"^^<someClass> supposed to mean? Are you punning on datatypes and classes?09:10
<acoburn>ajs6f: yes, exactly
<ajs6f>acoburn: No, I don't think you can do that. Let me check.
<acoburn>ajs6f: the w3c docs suggest that you can, let me find that...
ajs6f: https://www.w3.org/TR/rdf11-mt/#patterns-of-rdf-entailment-informative09:11
ajs6f: it's a non-normative part of the spec, but still...
<ajs6f>acoburn: That doesn't even seem vaguely on point. What are you trying to quote?
<acoburn>ajs6f: the example ex:a ex:p "123"^^xsd:integer09:12
* github-ff joins
[fcrepo4] awoods deleted FCREPO-1868 at 8204b18: https://git.io/vPlD1
* github-ff leaves
<ajs6f>acoburn: xsd:integer is not a class. It is a datatype. They are distinct.
<acoburn>ajs6f: the RDF entailment implies: ex:a ex:p _:x . _x: rdf:type xsd:integer .09:13
ajs6f: and isn't rdf:DataType a subset of rdf:Class?
<ajs6f>acoburn: I think I'm begging to see what is wrong here. This is why I was so irritated with the (foolish) decision to make untyped literals semantically equal to string-typed literals.09:14
<acoburn>ajs6f: my whole point here is trying to model some of these dcterms properties without needing hash URIs
<ajs6f>acoburn: This is now incompatible with OWL.
acoburn: That sucks.
acoburn: I don't understand what the braces in "RDF entails recognizing {xsd:integer}" mean. Do you?09:15
<acoburn>ajs6f: "rdfs:Datatype is both an instance of and a https://www.w3.org/TR/rdf-schema/#def-subclass ofhttps://www.w3.org/TR/rdf-schema/#ch_class"
<ajs6f>acoburn: That didn't come across (were there links in that c+p)?
<acoburn>ajs6f: "rdfs:Datatype is both an instance of and a subClass of Class"09:16
ajs6f: from https://www.w3.org/TR/rdf-schema/#ch_datatype
<ajs6f>acoburn: I don't understand what {xsd:integer} is.09:17
<acoburn>ajs6f: it seems from the example that it is both a datatype and a class09:18
ajs6f: that's what I'm trying to figure out
<ajs6f>acoburn: No, definitely not. It's not the same thing as xsd:integer no braces. It's a set. That seems to me to be a set satisfying interpretation.
* thomz leaves
<ajs6f>acoburn: RDF may be saying that xsd:integer is a class, but { xsd:integer } is clearly not.09:19
<acoburn>ajs6f: but what about the statement: _:x rdf:type xsd:integer ?09:20
<ajs6f>acoburn: I agree that that indicates that xsd:integer rdf:type rdfs:Class, which sounds like bullshit to me, but whatever. I'm really only interested in OWL.09:21
<acoburn>ajs6f: I was reading {xsd:integer} to mean the set of all entailed classes
<ajs6f>acoburn: That makes more sense.
<acoburn>ajs6f: basically, I want to know if this is valid RDF:
<ajs6f>acoburn: "Valid"?09:22
acoburn: You mean "satisfiable"?
acoburn: There is no "valid".
<acoburn>ajs6f: <resource> dcterms:extent "1 volume (248 pages)"^^dcterms:SizeOrDuration
ajs6f: I mean semantically satisfiable
ajs6f: actually, I mis-wrote that ^^^
<ajs6f>acoburn: I think it might be fine, and it's a great example of why this is bullshit. No one is going to take the time to figure out what you mean by that.09:23
<acoburn>ajs6f: <resource> dcterms:extent "1 volume (248 pages)"^^xsd:string
ajs6f: and whether that statement ^^^ implies:
<ajs6f>acoburn: See this is why this is dumb. The implication is "1 volume (248 pages)" rdf:type xsd:string, which is obvious wrong.
<acoburn>ajs6f: <resource> dcterms:extent [ rdf:value "1 volume (248 pages)"^^xsd:string ; rdf:type dcterms:SizeOrDuration ];09:24
<ajs6f>acoburn: You are really twisting yourself into knots over this.
acoburn: Do you actually plan to do inference over this data?
<acoburn>ajs6f: yes, I should probably just use hash URIs and be done with it
ajs6f: in the triplestore, yes
<ajs6f>acoburn: And is that going to be done with OWL?09:25
<acoburn>ajs6f: there is some simple OWL
<ajs6f>acoburn: then worry about _OWL_ semantics, which are independent of (and much superior to) RDF semantics.09:26
acoburn: None of these nonsense incantations arises.
<acoburn>ajs6f: fair enough, thanks!
<ajs6f>acoburn: Honestly, "1 volume (248 pages)" rdf:type xsd:string . So RDF now entails syntactically impossible results. Awesome. Well done, guys.09:27
acoburn: This is why _you start with logic_. DL, whatever fragment works for you.
<acoburn>ajs6f: I had been reading this as: "the node whose value is '1 volume (248 pages)' has a rdf:type of xsd:string"09:28
<ajs6f>acoburn: You are trying harder than the authors deserve.09:29
<acoburn>ajs6f: which is, admittedly, splitting hairs
<ajs6f>acoburn: Compare all this jazz to the simplicity and clarity of:09:31
and the power of:09:32
"Thus, datatypes are analogous to classes, the main difference being that the former contain data values such as strings and numbers, rather than individuals."
* github-ff joins
[fcrepo4] awoods pushed 1 new commit to master: https://git.io/vPl99
fcrepo4/master b59d251 Aaron Coburn: Add namespaces to the default CND configuration (#1109)...
* github-ff leaves
<ajs6f>How hard was that?
Oops, first.
fire_order_logic sounds like something used on a battleship.09:37
* mikeAtUVa joins
<ajs6f>RDF is jive.
* mohamedar1 joins09:52
* mohamedar leaves09:54
* travis-ci joins09:56
fcrepo4/fcrepo4#4733 (master - b59d251 : Aaron Coburn): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/8e7752d111c7...b59d2511bfb8
Build details : https://travis-ci.org/fcrepo4/fcrepo4/builds/165816277
* travis-ci leaves
* dhlamb joins10:08
<f4jenkins>Project fcrepo4 build #3404: UNSTABLE in 1 hr 5 min: http://jenkins.fcrepo.org/job/fcrepo4/3404/10:18
* mikeAtUVa leaves10:25
<f4jenkins>Yippee, build fixed!10:28
Project fcrepo4 build #3405: FIXED in 10 min: http://jenkins.fcrepo.org/job/fcrepo4/3405/
awoods: Add namespaces to the default CND configuration (#1109)
* mikeAtUVa joins
* mjgiarlo joins10:41
* dbernstein joins10:50
* dhlamb leaves10:52
* dhlamb joins10:54
* github-ff joins11:06
[fcrepo4] awoods pushed 1 new commit to master: https://git.io/vP8ep
fcrepo4/master 7816b5b Aaron Coburn: Make it possible to set the location for velocity.log (#1111)...
* github-ff leaves
* travis-ci joins11:29
fcrepo4/fcrepo4#4735 (master - 7816b5b : Aaron Coburn): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/b59d2511bfb8...7816b5b77a5b
Build details : https://travis-ci.org/fcrepo4/fcrepo4/builds/165844209
* travis-ci leaves
* bseeger leaves11:45
* bseeger joins12:26
<whikloj>acoburn++ # nice job cleaning up tickets13:17
<ajwagner>whikloj: I just succesfully ran tests against mysql/postgres, is there another priority test?13:37
<whikloj>ajwagner: Any that haven't been done at least once would be a priority, or anything you might be interested in13:38
<ajwagner>Outside of windows builds I'm only seeing vagrant without audit/auth that hasn't been tested.13:40
* acoburn leaves13:48
<whikloj>ajwagner: sure, unless you are in a position to attempt a migration of data
<ajwagner>whikloj: All of my data is back in Fedora 3 land, but I suppose I could spin up an earlier version import some sample data and then migrate...13:50
<whikloj>ajwagner: your call, but sample data makes it harder to tell if anything has changed not failed necessarily but is appearing differently after import.13:52
ajwagner: If you want to run any tests that someone else has already done, that is cool too.
ajwagner: thanks for the help
* github-ff joins13:57
[fcrepo4] acoburn opened pull request #1112: Add charset encoding for text/* content-type serializations (master...fcrepo-1750) https://git.io/vP8BN
* github-ff leaves
<ajwagner>whikloj: no problem, happy to help13:58
* youn leaves14:35
* acoburn joins14:52
* mikeAtUVa leaves14:56
* mikeAtUVa joins14:57
* kefo joins14:58
* ajs6f leaves15:08
* ajs6f joins15:09
* mohamedar1 leaves15:17
* kefo leaves15:18
* whikloj leaves
* kefo joins15:19
* whikloj joins
<mikeAtUVa>ajs6f: do you remember how to put content into fedora as a binary when (inexplicably in the case of text/csv) your content type is considered to be RDF?
<ajs6f>mikeAtUVa: You mean that _Fedora_ is somehow considering it to be RDF?
acoburn: ^^^ have you done anything to mimetype proccessing that might be connected to this?15:20
<mikeAtUVa>ajs6f: technically jean things it's RDF and we're taking their word for it.
<ajs6f>mikeAtUVa: jean? Jena?
<ajs6f>mikeAtUVa: What are you using? 4.7.0-RC?/15:21
<acoburn>ajs6f: no, but I recall something dchandekstark brought up about this a while back
<ajs6f>There are ways that Jena thinks about CSV as RDF. After all, that's basically what NTriples is.
But I thought we had our own independent lists of what we will accept as RDF?15:22
<mikeAtUVa>Yes, but it's been true since well before that.
<ajs6f>mikeAtUVa: What's been true?
* bseeger leaves15:23
ajs6f: our treatment of CSV.
<ajs6f>mikeAtUVa: but https://github.com/fcrepo4/fcrepo4/blob/master/fcrepo-http-api/src/main/java/org/fcrepo/http/api/FedoraLdp.java#L44715:25
We allow everything throught.
Yeah, org.apache.jena.riot.RDFLanguages.CSV15:26
That's what's doing it.15:27
We need to have an independent construction of the acceptable mimetypes for RDF. I thought we already did.15:29
acoburn: Didn't you have to do that for the JSON stuff, on the other (producing) end?15:30
<acoburn>ajs6f: in a mtg
<ajs6f>acoburn: Wow! You did all of that JSON stuff while you were in a meeting? Impressive!15:31
<mikeAtUVa>ajs6f: in the mean time, should I just lie about the mime type or do we have a standard practice when people want to shove RDF in as binary?
<ajs6f>mikeAtUVa: I dunno. What do you think?
mikeAtUVa: Do you think we need a "Prefer" header value of something like that?15:32
* mikeAtUVa does not really know.15:33
* dbernstein leaves
* westgard leaves
* anarchivist leaves
* dbernstein joins
* westgard joins
<mikeAtUVa>ajs6f: As a user (of fedora4) I'd always prefer something unambiguous even at the expense of being standards-based, so I'm a poor choice for doing design work. Anything I could include in the request to explicitly say to treat something as binary would be ideal.15:35
<ajs6f>mikeAtUVa: Standards? I don't see the connection...15:36
<whikloj>what about Content-type: application/octet-stream, then the data is untouched15:38
<mikeAtUVa>ajs6f: LDP has imposes some requirements on our interactions... which may rule out whatever naieve implementation I'd propose.
wikloj: sounds reasonable enough... for the sake of my application needs, the filename ending in .csv will be enough to tell future users how to interpret the binary.15:40
<ajs6f>mikeAtUVa: Sounds like a good workaround now. I don't see why we can't develop a more explicit technique for the future. Can you file a ticket?15:41
<whikloj>application/octet-stream is a good way to tell Fedora not to touch the data on the way in, but as ajs6f says its not ideal. Something to set the file mime-type separately would be awesome, but maybe that is just a question of a property on the fcr:metadata of the binary15:42
<mikeAtUVa>ajs6f: should I file a ticket for creating a way to specify that a PUT or POSTed resource is binary, despite its mime type, or one to allow text/csv to be treated as binary by default?15:43
<ajs6f>whikloj: A recording of mimetype as a property and the interaction model supported by the server are two different things
mikeAtUVa: The former. Let's do the general problem.
<whikloj>ajs6f: agreed, I'm just thinking about what mikeAtUVa said about letting future users know how to interpret the binary15:44
* anarchivist joins
<ajs6f>whikloj: Future users will have quantum computers. They won't care about this stuff.
<mikeAtUVa>ajs6f: I'm still tempted by the latter, because it's completely unexpected and wrecks havoc on your namespace registry (ie, adds a bunch of wierd stuff you can't delete)
<ajs6f>whikloj: They will use their quantum computers to interpret it all possible ways at once.
mikeAtUVa: If you think it's bad on its own account, file another ticket. I'm just saying we have two separable things here, not that we shouldn't deal with both.15:45
<mikeAtUVa>ajs6f: OK. tickets are cheap... I'll add two.
* f4jenkins joins23:13

Generated by Sualtam