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

Using timezone: Eastern Standard Time
* fcrepo-bot joins00:21
* fcrepo-bot leaves01:23
* jonathangee joins08:22
* ksclarke joins09:39
* edInCo joins10:24
* ajs6f joins10:29
awoods: ReadyTalk today? (Boo!)10:33
<awoods>ajs6f: yes, ReadyTalk
<ajs6f>awoods: Don't we have fifteen-person hangout now?10:34
<awoods>ajs6f: we do, and if we think for the foreseeable future 15 is enough, we can switch. But the constant Thursday confusion about which line to use argues for a consistent number... instead of flopping.10:35
<ajs6f>Hangout all the time would be consistent.
<awoods>if we think for the foreseeable future 15 is enough10:36
<ajs6f>awoods: I think so, but maybe we should just count the names on minutes pages.10:45
awoods: btw, how come you took my name off the contributing dev list for the "holiday release"? :(.
<awoods>ajs6f: ReadyTalk today, and we can discuss future plans on the call.10:46
ajs6f: which page?
<ajs6f>awoods: WHy not start with ReadyTalk and then switch to Hangout ten minutes in, w/o warning or explanation?
awoods: cmmorris' email.
afk bbs10:48
<awoods>ajs6f: my apologies. I have fixed the release notes: https://wiki.duraspace.org/display/FF/Fedora+4.0+Alpha+3+Release+Notes#Fedora4.0Alpha3ReleaseNotes-Acknowledgements10:51
* ajs6f1 joins10:53
* ajs6f leaves
* osmandin joins11:01
<ajs6f1>I'm here.
<edInCo>me too
<awoods>anyone else coming to ReadyTalk?11:03
<ajs6f1>Happy New Year to edInCo, but only six months to awoods. I'm holding back until I see performance.
* mikeAtUVa joins11:06
<ajs6f1>If that's an issue, we can assign a person who has a phone,11:07
to be the registered "phone-in" person.
GNU Asterisk works.11:08
Money we _could_ be spending on beer.11:10
cbeer++ and mikeAtUVa++ for helping make the release happen.11:21
<barmintor>I'm sure I'll be able to negotiate time to work on JAXRS211:22
(no mic)11:23
<ajs6f1>barmintor needs no mic for Jersey2.
<barmintor>yes, I'm on the line
we should leave it fornow
<ajs6f1>We can get rid of the web.xml stuff via:11:26
As long as we are bound by MODE's JSON, there's nothing we can do about that.11:27
That's why we had that conv with rhauch about programmatic config for MODE.
That would allow us to factor that confi out into a space we control.
<osmandin>There's also pom.xml change11:28
<ajs6f1>osmandin: That's where we need a modularization frameowkr, like OSGi or something like it.
awoods: It would get rid of web.xml, periiod.11:29
osmandin: Right now our mod frameowkr is… Maven.
Which is not a good choice.
I think cbeer is.11:32
<barmintor>awoods: it's not impossible that Rob commands me to go
<ajs6f1>Cartolano COMMANDS!11:33
awoods: They're going to want to talk about RDF a lot.11:34
awoods: I think you should just wave your hand a lot.
awoods: cbeer knows more than I do.
ask him.
Declan will say things about blank nodes.11:35
Put a beer in his hand to calm him down.
<barmintor>that TX problem sounds like an issue with thread-local state, or TXMananger bean binding11:37
<ajs6f1>barmintor: When you tried Jersey2, you ended up using a JAX-RS Application class, right?
<barmintor>ajs6f1: Yes
<ajs6f1>barmintor: I didn't like that then, but I ;ike it now.
We can use Servlet 3.0 annotations that way.11:38
<barmintor>ajs6f1: I'm ambivalent- depends on whether the scanning can work. I don't like registering components there
<ajs6f1>awoods: Do we have a requirements-list for xactions?
barmintor: That's what I didn't like.
<barmintor>but there's a lot of work and debugging going on in the Jersey 2.x line right now
<ajs6f1>barmintor: I wouldn't mind a simple but flexible Application that just picks up some metadata indecating how to scan.11:39
Like it could select for some types to scan from a service registry, like OSGi or Spring.
<barmintor>sure. that sounds great.11:41
<ajs6f1>barmintor: Okay, I'll get ready to merge it when you send the PR>11:46
awoods: I see cmmorris put my name back in. Thanks. ;)11:49
<awoods>ajs6f1, yes. Sorry about that.
<ajs6f1>awoods: It's okay. I'm used to giving and giving and giving, only to be slapped in the face. It's what being a Fedora committer is all about.11:50
<awoods>ajs6f1, I appreciate your comedy, but will refrain from commenting.11:51
<ajs6f1>Too much love is worse than none at all.11:52
<awoods>ajs6f1, is it safe to conclude that our notion of the wiring/config capability depends on the presence of Jersey2?11:56
<ajs6f1>I was looking at the code4lib presentation voting— escowles proffer is the 3rd highest. That's pretty cool.
awoods: No, but it would be good to tackle J2 first.
awoods: It would tell us about what committments we might have to CDI or HK2.
<awoods>ajs6f1, ok.11:59
<ajs6f1>awoods: If Jersey commits us to engaging HK2 (or CDI), it's not so much that it make those choices any better— it does make other choices more expensive.
<awoods>ajs6f1, sure. It makes sense to see what the F4/J2 picture looks like so we can make a further informed decision about how to best leverage that framework and its cronies for the wiring/config.12:01
<ajs6f1>awoods: Or, we could blast ahead on a nitro-fueled code funnycar into an unintelligible world of opaque declarative config and endless layers of typed contracts for wiring.12:02
That's what I'd rather do.
<awoods>ajs6f1, we all appreciate your restraint.12:03
afk bbl12:11
* ajs6f1 leaves
* jonathangee leaves14:56
* edInCo leaves15:27
* ajs6f joins15:46
all: fcrepo-jcr has an explicit dependency to commons-codec 1.6. I can't see any earthly reason for it— does anyone know why it's there?15:47
<barmintor>Probably I did it, for a reason.15:48
<ajs6f>barmintor: Problem sovled!15:49
Seriously, any idea? The single class in fcrepo-jcr doesn't use it.
Trying to juggle dependency versions?
<barmintor>I was being serious.
<ajs6f>I know, but I need even more serious help. Or I can just send a PR with it removed.15:50
<barmintor>I think it was to override the version of an excluded transitive dependency15:51
but why there/
<ajs6f>Exactly— why there of all places?15:52
<barmintor>Related: Why does that module have any library dependencies at all?15:53
Including JBoss JTA, LevelDB stuff, something called "Snappy Java"?15:54
<barmintor>Is it to somehow peg versions, sinc fcrepo-jcr is itself a dependency of other modules
especially since client modules of fcrepo-jcr may not be child modules of fcrepo4?
<ajs6f>Something called "vorbis-java-tika", which sounds like either a monster from Lovecraft or a delicious fruit-juice-based cocktail.15:55
barmintor: What would depend on fcrepo-jcr directly?
<barmintor>modeshape stuff (like listeners) that use those constants but don't depend on any given fcrepo4 module
<ajs6f>barmintor: And if something does, than we have a new ticket: make a BOM for fcrepo so that people don't depend on inappropriate modules for locking versions of deps.15:56
barmintor: Okay, but ^^^
barmintor: Oh, way, the constants.
<barmintor>ajs6f1: Fine, but vvv
<ajs6f>Yeah, just got it.
Hm. Maybe factor out a dep-less artifact with constants… fcrepo-dictionary-of-incantation-names.15:57
We're crossing concerns here, and there's secretly two artifacts here. One: a dictionary of useful constants, two: a dictionary of correct dependency versions.15:58
<barmintor>in addition to the constants, there are also the example MODE configs16:03
<ajs6f>barmintor: Right. Is the source code even the right place for them? Are we using them in tests? (All of them?) Should they just be in the wiki somewhere?16:04
<barmintor>Maybe, Yes, (I think most of them), That's probably insufficient16:05
<ajs6f>Meaning a wiki place isn't sufficient?
<barmintor>not in my accounting16:06
<ajs6f>Okay, maybe a "fcrepo-example-mode-configs"?
<barmintor>so many dependencies...
<ajs6f>barmintor: I don't like multiplying entities either, but I like even less when we cross concerns and then have to separate them later at greater cost.
<barmintor>ajs6f: This sounds like a job for the TECH LEAD *krakow!*
(not Poland)16:08
<ajs6f>barmintor: and I just tried to start an experimental indexer and ran right away into version problems because of the fcrepo-jcr stuff.
Now I will always picture awoods in a superhero costume with "TL" logo and holding a golonka in one hand.16:09
awoods: I'll wrrite you some tix.16:10
<pivotal-bot>A. "Horbulaco" Soroka added "Separate example MODE config catalog into a new artifact" https://www.pivotaltracker.com/story/show/6317246016:12
A. "Horbulaco" Soroka added "Separate Java constants now found in fcrepo-jcr into a new artifact" https://www.pivotaltracker.com/story/show/63172482
<ajs6f>Well, if I was a dev aspiring to write a new indexer, I would pretty much stop now. So I guess I will. Back to real life.16:13
* ajs6f leaves
<awoods>barmintor: I do not think there are any MODE config examples in fcrepo-jcr. I am misreading this?16:24
<pivotal-bot>Andrew Woods added comment: "Which MODE config catalog do you have in mind, @ajs6f?" https://www.pivotaltracker.com/story/show/6317246016:32
Andrew Woods added comment: "The Java constants are all that is in fcrepo-jcr. It seems odd to move them to another artifact which will ..." https://www.pivotaltracker.com/story/show/6317248216:34
* mikeAtUVa leaves16:39
* osmandin leaves16:45
<pivotal-bot>Andrew Woods added "Only enable AuthZ UI form when applicable" https://www.pivotaltracker.com/story/show/6317822817:44
* mikeAtUVa joins18:09
* ksclarke leaves18:10
* mikeAtUVa leaves18:11
* ksclarke joins19:09
* fcrepo-bot joins22:55