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

Using timezone: Eastern Standard Time
* dchandekstark joins01:02
* dchandekstark leaves01:07
* mjgiarlo joins01:29
* mjgiarlo leaves01:50
* mjgiarlo1 joins
* thomz joins02:43
* tjohnson leaves02:57
* dchandekstark joins07:37
* dchandekstark leaves
* dchandekstark joins
* dchandekstark leaves07:38
* dchandekstark joins
* dchandekstark leaves07:39
* peichman joins07:54
* f4jenkins leaves07:56
* peichman leaves07:58
* f4jenkins joins08:02
* dchandekstark joins08:26
* dchandekstark leaves08:31
* dwilcox joins08:35
* mikeAtUVa joins09:00
* bseeger joins09:03
* acoburn joins09:08
* whikloj joins09:15
* dchandekstark joins09:18
* dchandek_ joins09:20
* dchandekstark leaves09:22
* dchandek_ leaves09:24
* dchandekstark joins
* dwilcox leaves10:02
* thomz leaves10:07
* bseeger leaves10:08
* ajs6f joins10:20
acoburn:ruebot:whikloj:barmintor: You want to run the meeting, or you want me to, or should we make someone else do it?10:25
* dwilcox joins10:27
<acoburn>ajs6f: let's ask for volunteers at 1110:32
<ajs6f>acoburn: Okelely dokeley.
* dchandekstark leaves10:35
* dchandekstark joins10:42
<ajs6f>acoburn: awoods hates the idea of a slipped build so much that he offered to rebase your Jena PR as needed.10:45
<acoburn>ajs6f: I saw that. we need to figure out what to do with fcrepo-transform10:46
<ajs6f>acoburn: We have an official policy: dissolve it into fcrepo-camel.
<acoburn>ajs6f: I'm 100% in favor of that
* dchandekstark leaves10:47
<acoburn>ajs6f: in principle, camel isn't even necessary, we can create an OSGi artifact out of it and just run it in karaf as a service
<ajs6f>acoburn: awoods wanted to be sure that all use cases were covered. I don't think he took into account the fact that we have no official use cases for fcrepo-transform. We just threw it out there.
acoburn: Te only thing I know of anyone doing with it is indexing.
<acoburn>ajs6f: I find it useful, but it would be _really_ useful if it used the linkeddata backend cache10:48
<ajs6f>acoburn: Sure, we can wrap it however we like, but proxying web endpoints with a transforming pipeline, that sure sounds like Camel.
<acoburn>ajs6f: otherwise, it's just a not very useful toy
<ajs6f>acoburn: Yes, that's why I filed the ticket. But it makes more sense to move that outside the core repository process.
<acoburn>ajs6f: it really sounds like API-X to me
<ajs6f>acoburn: We should do it, just not inside the repository.
<ajs6f>acoburn: I would love to use API-X for it, but I don't have a sense that API-X is ready to do it.
acoburn: Maybe API-X is where this eventually ends up.
<acoburn>ajs6f: I agree, but structuring it in such a way that API-X can use it makes sense to me
ajs6f: incidentally, I have a branch _right now_ that has the linked data cache implemented with fcrepo-transform10:50
<ajs6f>acoburn: I'm in no ay opposed, but how do we do that? Do we know enough to do that now?
<acoburn>ajs6f: I've been implementing services at Amherst that are all intended to fit into API-X
<ajs6f>acoburn: Wait, you packed Marmotta inside fcrepo-transform— did you use the fact that Jena's namsepace changed to be orthagonal to itself?10:51
<acoburn>ajs6f: yep
<ajs6f>acoburn: I did that for trippi-sparql.
acoburn: It's goofy akido, but it works.
<acoburn>ajs6f: no, I didn't do that — I removed the jena backend
ajs6f: and replaced it with the linkeddata backend
<ajs6f>acoburn: I'm talking about having both old-jena dna new-jena in the classpath.10:52
acobun: Doesnt the LD backend require old-jena?
<acoburn>ajs6f: right — that's not what I did
ajs6f: no, it requires old commons-httpclient (another issue)
<ajs6f>acoburn: Wait, LDCache doesn't require old-jena?
<acoburn>ajs6f: no10:53
ajs6f: it uses sesame
<ajs6f>acoburn; I could have sworn… ow well.
acoburn: So what is awoods' hangup?
acoburn: We have branches for both sides?
<acoburn>ajs6f: at this point, probably that the maven enforcer doesn't like commons-httpclient
<ajs6f>acoburn: Oh, for crying out loud.10:54
acoburn: is it 3->4?
<acoburn>ajs6f: marmotta-commons uses v3
<ajs6f>acoburn: Yeah, I'm still moving Jena to 4. It's a jump. And Jena is far more responsive in terms of releasing than Marmotta.
<acoburn>ajs6f: as does sesame-http-client
<ajs6f>acoburn: Wait, what version of Sesame are they sing?10:55
<acoburn>ajs6f: 2.7.11
<ajs6f>acoburn: Urp.
acoburn: They haven't even gotten to teh 3-serie yet.
<acoburn>ajs6f: but I do a bunch of terribly ugly maven gymnastics to avoid collisions with the existing 2.7.13 required by fcrepo-http-commons10:56
acoburn; Why do we have Sesame in the classpath at all?
<acoburn>ajs6f: I *think* because it handles streaming RDF output, but I'd like to review that
ajs6f: I don't know why jena can't do that10:57
<ajs6f>acoburn: That's a terrible reason. We can do that easily with jena.
<acoburn>ajs6f: as you know, I am always in favor of reducing/simplifying dependencies
<ajs6f>acoburn: Let's get rid of MODE, then.10:58
<acoburn>ajs6f: at least JCR
<barmintor>acoburn: trying to decompose URI-Node into URI-(Path Abstraction and (Path Abstraction-Node) to try to enable MODE querying instead of node iteraction
<ajs6f>ruebot: You have a loud background.
ruebot: I'm a big fan of ruebot.
<barmintor>acoburn: and seeing why you recoiled in horror a while back
<ruebot>ajs6f: you gotta get the "I TRUST RUEBOT" t-shirt
<barmintor>especially since we don't have a path abstraction10:59
<ajs6f>barmintor: Wouldn't you rather start afresh? Doesn't that seem more and more tempting with every day?
* escowles joins
<ajs6f>ruebot: Where is your Cafe Press shop?
<acoburn>barmintor: as I recall, there is a lot (too much?) complexity in the code there right now. I'm not surprised that it may be slow
* kat3_drx joins
* ruebot looks at his myspace for the link
<barmintor>ajs6f: it depends on whether I'm thinking about the folks who moved before me, but I can't in good conscience entertain the idea right now11:00
<acoburn>*is here*
<ajs6f>barmintor: NP: I'll think about it enough for both of us. Then I'll start work on it.
* ajs6f is here
* apb18 joins11:01
* barmintor is here
* mikeAtUVa is here
* escowles is here11:02
* apb18 is here
<barmintor>RUEBOT FOREVER11:03
<ajs6f>I trust ruebot.
<whikloj>I trust no one, but if I did, I'd trust ruebot
<ajs6f>We need more scale in our perf/scale group.
dwilcox: is that you or a panda typing?11:08
<ajs6f>apb18: On the release testing page for 4.6.0, did you forget to put your name next to the successful test for fcrepo-webapp-plus "mvn clean install -P\!webac,\!audit" on Windows? Was that you?
<dwilcox>ruebot: Must be a panda - I am muted
<barmintor>moving the performance problem to items in many collections creates the perverse scenario where your most frequently linked objects are the slowest :(
<apb18>ajs6f: oh, yes. That was me
<ajs6f>barmintor: Not if you are hitting indexes and not the repository.11:09
<ajs6f>apb18: Thanks for all that Wintesting!
* bseeger joins11:10
<ajs6f>acoburn has been trying to get that on the agenca for weeks.
* dchandekstark joins
<ajs6f>Is this something you can do at the container level?
<bseeger>*is here*11:11
<apb18>ajs6f: No problem, seeing things work on Windows satisfies a sick sense of curiosity.11:12
<ajs6f>apb18: Yeah, it's like watching a monkey make sushi.11:13
<escowles>fwiw, tomcat does support gzip compression at the container level, with some options for size/mime type, etc.: https://tomcat.apache.org/tomcat-8.0-doc/config/http.html11:14
* dchandekstark leaves11:15
<ajs6f>escowles: The only thing about that is that we don't want to get into teaching people how to do container config. We did too much of that for Fedora3.
<escowles>jetty has something similar: http://www.eclipse.org/jetty/documentation/9.4.x/gzip-filter.html11:16
* dchandekstark joins
<ajs6f>escowles: I bet pretty much any container will. It's clearly a natural concern for containers. But it seems pretty cheap to buy it at the app level.
<escowles>ajs6f: i completely agree we don't want to get into documenting and supporting this — but we could reasonably say that if you want gzip compression, that your container could do it for you
<ajs6f>escowles: Yeah, I'm not on fire to see it done at the app level.11:17
escowles: Do you actually object to doing the Jersey filter thing?11:19
<escowles>ajs6f: no, i don't object — as long as it can be implemented without complicating the LDP implementation, i'm fine with it11:20
<ajs6f>And these are all things that are completely central to indexing.
And other things.
* dchandekstark leaves11:21
<ajs6f>escowles; I shouldn't think that they would interact, right?
escowles: Maybe let's file the ticket, and see what comes of it.11:22
<escowles>ajs6f: yes, i 'jersey filters' sounds like it would be completely isolated
<ajs6f>escowles: Jersey filters sounds even lamer than Jersey shore.
<ruebot>"in an uber-jar"
<barmintor>which is something we can relate to11:24
<ajs6f>barmintor: Letting PRs sit?
<escowles>being a small group that sometimes moves slowly11:25
<barmintor>small team, limited velocity
<ajs6f>barmintor:escowles: Fedora Army is endless.
<apb18>Jena streaming works well, in my experience11:29
<escowles>+1 - i can't imagine a reason we'd need sesame and jena both
<ajs6f>escowles: Because we're lonely.
<ruebot>ajs6f: point me to it, and i'll do it this afternoon. with linux && firefox11:31
<apb18>The Edge browser seems to work with the UI for the first time ever
I can see about fitting in the manual tests for Edge, possibly tomorrow.11:33
<ruebot>Bono 2.0
<apb18>Running ITs against externally-configured fedora worked reasonably well for Fedora 2/3 testing
<ruebot>ajs6f: ping me for vagrant.11:37
<whikloj>ruebot: ajs6f: I'm running some vagrant ups now
<barmintor>chip chip chip
tink tink tink
<ajs6f>Ooh, sing that 7 dwarves song from Disney's Snow White!@
"Whistle While You Work"
Fix all the fixities
<barmintor>here's an fyi: while I work on the property-uri resolution bit, I'm also trying to document the closely related thinking behind the fcr:content/fcr:metadata question11:41
<ajs6f>barmintor: Oh, don't tel me we're gong to change it back again.
<barmintor>ajs6f: no, but I want a thing tp point to
instead of shrugging and saying "it's been both ways I guess"11:42
<ajs6f>barmintor: How about a GIF of my tired, tired face.
<barmintor>which is what I usually do
<ajs6f>barmintor: I can't remember why we didn't do it with a describedBy header.
<barmintor>ajs6f: because the file properties wanted to be deleted with the binary
ajs6f: and fcr:metadata seemed like a way to do it11:44
<ajs6f>barmintor: But that's an impl concern. We can record stuff however we want, but how we expose it is aff question.
<ajs6f>barmintor: The more I thin about it, the more I really wonder why we need fcr:metadata.11:45
<barmintor>maybe a bbetter way to say it is that we understand LDP better now than we did in September of 2014
<ajs6f>barmintor: Yes, and RESTful style. But what do we do with that new wisdom and fu?
* mjgiarlo joins
<barmintor>ajs6f: weep
<ajs6f>barmintor: Seriously, we need to get the spec right. I don't want to hurt anyone using the software out now, but I _really_ don't want to bake junk into the specs.
<barmintor>ajs6f: fcr:metadata is definitely not in the specs11:47
<ajs6f>barmintor: But do we have a solid pattern for description that is? Do we just adopt LDPs MAY as a MUST?
<barmintor>ajs6f: MUST
<ajs6f>barmintor: And we really think that current clients aren't doing x +"/fcr:metadata" instead of following the link?11:48
<barmintor>ajs6f: I don't know, but ActiveFedora follows the link AFAIK11:49
<ajs6f>barmintor: Well, maybe I'm worrying too much.
<acoburn>ajs6f: fcrepo-camel and fcrepo-java-client both follow the link
<ajs6f>barmintor: You know, I'd almost want to go through the current API docs and actually remove references to fcr:metadata and just talk about the link.
barmintor: The fact of the URL pattern is not part of the API. It's historical impl.
<apb18>ruebot: is there a import/export meeting tomorrow?11:53
* kat3_drx leaves11:55
<apb18>ruebot: Thanks!
<ruebot>apb18: de nada!11:56
<ajs6f>tagging out RC-312:00
* github-ff joins
[fcrepo4] ajs6f tagged fcrepo4-4.6.0-RC-3 at 1fc2954: https://git.io/vKA6v
* github-ff leaves
<ruebot>ajs6f: let me know when the single-click 4.6.0-RC-3 is ready, and i'll test that on linux, and poke through the interface on firefox too.12:03
Just looking at that now.
<ajs6f>Oh, phooey. Github automade a release when I tagged. Now I have to figure out how to add artifacts to that.
<ruebot>ajs6f: oh, that's easy
" Attach binaries by dropping them here or selecting them."12:05
* github-ff joins12:10
[fcrepo-webapp-plus] acoburn opened pull request #43: Update enforcer rules to allow marmotta's use of commons-httpclient (master...fcrepo-1672) https://git.io/vKAiM
* github-ff leaves
* bseeger leaves12:12
<barmintor>the other option would have been separate jcr nodes for fcr:content && fcr:metadata, I guess
and a parent node for the NonRdfSource
<ruebot>dwilcox: let me know when you're done with the notes, and i'll send a note out to the list awoods style.12:13
<dwilcox>ruebot ++12:15
<ruebot>acoburn, ajs6f: https://wiki.duraspace.org/display/FF/2016-07-28+-+Fedora+Tech+Meeting -- can you make sure the proposals are captured correctly? and i'll send a note out.12:16
<ajs6f>ruebOt: That all seems plausible to me. I'm not that interested in the idea of a coherent shared history, anyway.12:20
ruebot: https://github.com/fcrepo4/fcrepo4/releases/tag/fcrepo4-4.6.0-RC-3
* dchandekstark joins12:23
<ruebot>ajs6f: where is the list of things i need to do in the UI?12:27
<ajs6f>ruebot: https://wiki.duraspace.org/display/FF/Release+Testing+-+4.6.0#ReleaseTesting-4.6.0-ManualTests
no one likes doing the manual tests, so thanks extra. I wonder if we can automate this browser stuff. I know people do automate this kind of thing all the time.
* dchandekstark leaves
* travis-ci joins12:30
fcrepo4/fcrepo4#4632 (fcrepo4-4.6.0-RC-3 - 0c3ff77 : birkland): The build passed.
Change view : https://github.com/fcrepo4/fcrepo4/compare/fcrepo4-4.6.0-RC-3
Build details : https://travis-ci.org/fcrepo4/fcrepo4/builds/148077045
* travis-ci leaves
* dchandekstark joins12:37
<ajs6f>afk bbl12:41
* ajs6f leaves
* tjohnson joins13:07
* bseeger joins13:21
* apb18 leaves13:54
* bseeger leaves14:09
* acoburn leaves14:10
* ajs6f joins14:41
* acoburn joins14:45
ajs6f: btw, it turns out we don't need to make changes to the maven-enforcer plugin; I can just exclude the commons-httpclient altogether14:52
<ajs6f>acoburn: via <exclusions> in the dependencies?
<acoburn>ajs6f: yes14:53
<ajs6f>acoburn: And it works? LDCche isn't using the old Commons 3 API?
<acoburn>ajs6f: that is correct
<ajs6f>acoburn: SWEET!
acoburn: will you close that PR then?
<acoburn>ajs6f: the only trick now is connecting in auth
ajs6f: yes, I'll do that
<ajs6f>acoburn: Do we have to have auth to roll? How many linked data sites have auth?14:54
acoburn: Mostly people will be hitting like id.loc.gov or VIA or the like.
<acoburn>ajs6f: the issue is when Fedora has auth enabled
<ajs6f>acoburn: Oh, if you're going back against your own repo.
<acoburn>ajs6f: right, and that's where fcr:transform starts
ajs6f: I'm assuming that any other auth is out of scope14:55
ajs6f: at that point, someone needs to wire in their own custom "provider"
<ajs6f>acoburn: And Marmotta doesn't seem to contemplate that at all, so it'll have to be somehow via the Commons API. You'll have to set the default Httpontext.
Damnin HttpContext.
<acoburn>ajs6f: right, I can set the HttpClient, so I'm working with that14:56
<ajs6f>acoburn: Oh, cool, that works.
<acoburn>ajs6f: basically, inject a Credentials (optionally) and an AuthScope, and if those are set, wire it all up
<ajs6f>acoburn: Sounds good to me. I'd be a little wondery about more complicated auth regimes, but lets get something out there so people understand how cool the tool is.14:57
<acoburn>ajs6f: yes, it's actually _extremely_ cool
<ajs6f>Once we're inside their guard, we'll stab fast and twist the knife.
I mean, we'll continue to help them accomplish their work!14:58
That's what I meant.
* bseeger joins15:04
* acoburn leaves
<ajs6f>acoburn:are you setting this thing up for an in-memory cache only, or are you peristing it to disk?15:12
* dchandekstark leaves15:14
* dchandekstark joins15:18
* dwilcox leaves15:23
* bseeger leaves15:24
* dwilcox joins15:26
<whikloj>ajs6f: are you making RC-3 wars for fcrepo-webapp-plus?15:52
<ajs6f> whikloj: Shoot— thanks for catching that, I'll do it now.
<whikloj>ajs6f: I'm out of town for the weekend, but I'll get those vagrant tests run on Monday. Or ruebot might beat me too it.
<ajs6f>whikloj: Enjoy your out-of-towness.
<whikloj>ajs6f: I always do
* github-ff joins15:56
[fcrepo-webapp-plus] ajs6f tagged fcrepo-webapp-plus-4.6.0-RC-3 at 989406e: https://git.io/vKxGs
* github-ff leaves
<ajs6f>whikloj: Just uploaded that.15:57
<ajs6f>whikloj: Thatnks for catching it!
escowles: Thank you for doing those tests with the SL db configs!16:02
ajwagner: Thatnks for adding in those test results!16:03
<escowles>ajs6f: no problem
* travis-ci joins16:07
fcrepo4-exts/fcrepo-webapp-plus#145 (fcrepo-webapp-plus-4.6.0-RC-3 - f89dbc9 : Aaron Birkland): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-webapp-plus/compare/fcrepo-webapp-plus-4.6.0-RC-3
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-webapp-plus/builds/148134133
* travis-ci leaves
<whikloj>ajs6f: I'm going to generate the WAR files for fcrepo-webapp-plus, cause I need them for vagrant. Do you think it would be okay to use the RC-2 for fcrepo-transform?16:08
<ajs6f>whikloj: Wait, why wouldn't you use the RCs… do you have to use something else? (I obviously don't know much about vagrant.)16:09
<whikloj>ajs6f: part of the vagrant startup process is to download the fcrepo-webapp-plus WARs from github.
ajs6f: So a valid test of just bringing it up means they have to be build and uploaded.16:10
<ajs6f>whikloj: Wait, are you saying that you want to put an actual 4.6.0 release on Github?
<whikloj>ajs6f: no
<ajs6f>whikloj: Yeah, I just put those there today. I'm not sure now what you want to do...?16:11
<whikloj>ajs6f: sorry like awoods did here
ajs6f: but under the new tag16:12
<ajs6f>whikloj: So you want to put the various flavors of fcrepo-webapp-plus up there?
<whikloj>ajs6f: yep
<ajs6f>whikloj: Ah, okay, that's cool. Yeah, there should be no issue with the fcrepo-transform module, I don't think. Are you worried about something in particular?16:13
<whikloj>ajs6f: no, nothing changes in it. So I figured I had built it with RC-2 which was the same as RC-1....so why do it again
<ajs6f>whikloj: Yeah, I don't see any problem with that. I didn't actually put a new tag in fcrepo-transform for that reason.16:14
whikloj: I think introducing a new tag is more confusing that otherwise.
<whikloj>ajs6f: yes I agree16:15
ajs6f: hmmm all the pom files in fcrepo-webapp-plus are still 4.5.2-SNAPSHOT, is that right or should they be 4.6.0-SNAPSHOT?16:16
<ajs6f>whikloj: Sigh. No, that is correct. It's very confusing and we are going to fix that next release, but it correct for now.16:17
<whikloj>ajs6f: ok cool
* bseeger joins16:24
* dwilcox leaves16:27
<ajwagner>ajs6f: no problem, happy to run more tests tomorrow if there is still a need16:34
<ajs6f>out for the day16:37
* ajs6f leaves
* dchandekstark leaves16:46
* bseeger leaves
* cmmills leaves16:47
* mikeAtUVa leaves16:58
* whikloj leaves16:59
* dchandekstark joins17:00
* dchandekstark leaves18:04
* dchandekstark joins18:09
* dchandekstark leaves18:14
* mjgiarlo leaves19:05
* dwilcox joins19:17
* dwilcox leaves19:35
* dwilcox joins20:02
* dwilcox leaves20:09
* dchandekstark joins21:08
* dchandekstark leaves23:17
* dchandekstark joins23:45
* dchandekstark leaves23:50

Generated by Sualtam