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

Using timezone: Eastern Standard Time
* eddies leaves01:43
* eddies joins01:44
* eddies leaves
* eddies joins
* eddies leaves02:46
* JasonDGI joins07:53
is the meeting today or tomorrow?09:18
<cbeer>JasonDGI: i thought today.10:03
<JasonDGI>i say armintor's update on ff-internal that made it sound like we're not starting until tomorrow, so you dont think thats the case?10:06
<cbeer>hm. i'm not seeing an email yet.10:10
<cbeer>eddies and jonathangee's emails seem to suggest a call today, though
<JasonDGI>is what i saw
cool, i can dig that
<cbeer>hm. where did the TT meetings page go?10:57
* eddies joins10:58
* eddies leaves
* eddies joins
<cbeer>oh, to the Fedora Futures section
just looking for the dial-in info
<eddies>i have a steering group call that overlaps, so i'll have to be jump off pretty quick11:00
* barmintor joins11:03
* barmintor leaves
* barmintor joins
<cbeer>eddies: added people to ff-tech and ff-internal lists; will create AWS credentials for team.11:06
barmintor: http://ffirc.cbeer.info/2013-01-07.html11:08
<barmintor>I'm going to email Neil & Eddie about attendance at c4l11:10
* ajs6f joins11:12
<cbeer>ajs6f: git://github.com/ajs6f/ff-modeshape-jmeter.git, right?11:13
<cbeer>sorry, the new github interface is confusing me
<ajs6f>It's huge, because I packed a preconfigured Jetty with ModeShape into it
to get you up and running quick.
But you can easily see where the script itself is.
and here:
is the API documentation.11:14
Right now, I'm a little unimpressed— I'm seeing occasional 500s thrown under concurrent load. I'm not sure whether this is because
I'm using the out-of-the-box configuration and that's not intended, or11:15
whether there is just some real weakness in the ModeShape HTTP API component.
<cbeer>the weird thing about modeshape for me (and i think you captured it on the wiki already) is it seems like we'll end up working on both sides of modeshape, both really low-level persistence plugins and some kind of Fedora-bespoke API on top11:20
<ajs6f>I think that is a consequence of our ambit: we care about durability _and_ operational elegance.11:21
I agree that MS wouldn't be a "black box".
The hope would be that its persistence design is a good framework for us to make the guarantees we want to make, and11:22
<cbeer>i don't find that a problem though. i think trying to treat anything as black-box in the long run is.... optimistic
<ajs6f>the upper level design is also a good place to work.
Yeah, I agree. If the problems we're trying to solve could be solved by assembling boxes, that's what we would have done. {grin}
I do think that the JCR abstractions around relationships vs. content are good. Not perfect, but good.11:23
<barmintor>Guh, I wish Github would display actual dates for tags/commits rather than "X months ago"
<ajs6f>And some of the questions I have around DataBank concern its treatment of relationships. It stores them, but that's about it.11:24
Ben— you can remember actual dates you did stuff? I can barely remember what I was doing last week.
<barmintor>ajs6f: No, I can't. That's why I'd like them displayed on Github :P11:25
<ajs6f>Ahh… you want a distributed brain.
<barmintor>(manager wants a summary of my work on Fedora during the 2011-12 fiscal year)11:26
<cbeer>git log --author=barmintor11:27
<ajs6f>You can add, "Ben made Adam Soroka's life a lot easier and better."
<cbeer>git log --author=barmintor --since=2011 --until=2012
<ajs6f>That should impress him.
<barmintor>cbeer: thanks
<cbeer>that's not quite right, i'm sure. author is probably an email or something. whatever your git name is11:28
ah, cool, it's actually a regex pattern11:29
* ajs6f leaves
* ajs6f joins11:40
* eddies1 joins11:54
* eddies1 leaves
* eddies1 joins
* eddies leaves
<barmintor>afk a second12:01
<eddies1>fyi: i'm off my calls12:02
<barmintor>eddies1: do we want to talk about c4l?12:04
<barmintor>let me hop back on the call
oh, skype is fine12:05
<eddies1>oh, i just hopped on the line
this way, others can join too ;-)
<barmintor>eddies1: https://maps.google.com/maps?saddr=The+Crowne+Plaza+Chicago+Metro+Hotel,+West+Madison+Street,+Chicago,+IL&daddr=University+of+Illinois+-+Chicago,+Chicago,+IL&hl=en&sll=41.868858,-87.673388&sspn=0.051836,0.078192&geocode=FQANfwIdWp3G-iFIBc_3daO8wClNuVWWxCwOiDFIBc_3daO8wA%3BFQPlfgId7zHG-iFoSJTYxcyIPylfmPRs5CwOiDFoSJTYxcyIPw&oq=733&mra=ls&t=m&z=1512:06
<cbeer>eddies1: another option suggested to me over break is vrbo.com, seemed to have some big houses too12:12
<eddies1>oh i had forgotten about that one. i used it years ago for a vacation
<ajs6f>Just got dropped. grr..12:20
<cbeer>Minimum stay requirements are 6 months12:23
ok, as far as airbnb goes, I think https://www.airbnb.com.sg/rooms/699284 and https://www.airbnb.com.sg/rooms/89385 are the only viable options (distance & location)13:02
<cbeer>eddies1: no wi-fi listed on 699284?13:56
i guess one of us should just remember to pack a router?
<eddies1>but it's listed in the description, i think the owner just didnt check the box13:57
<cbeer>oh, but they list it in the narrative
<barmintor>watch out: he's packing a router.14:16
* ajs6f leaves15:17
* JasonDGI leaves15:37
<cbeer>ok, time to work on ff stuff again. looking at ajs6f's thing..15:56
i was able to run ajs6f's test case successfully, didn't even see the 500s he was seeing16:20
and i'm glad he knows how to set up jmeter tests16:21
barmintor: ping?16:38
<barmintor>cbeer: heya17:14
<cbeer>barmintor: uh. i had a good question to ask you.
barmintor: oh, jmeter.
so, ajs6f has a simple set of jmeter tests for the modeshape api
<barmintor>I used it once.
<cbeer>and i was wondering, do you know if there's one for fcrepo?17:15
i mean, i assume people have done lots (ha) of one-off testing
<barmintor>there is not.
<cbeer>ok. so maybe it'd be valuable to know how slow fedora is, to compare it to the modeshape + databank numbers?
<barmintor>we worked on one, but didn't get it into useful shape, during the 3.6 development process
yes, quite valuable17:16
<cbeer>ok. i can copy/paste from ajs6f's stuff easily enough. it just does simple CRUD operations (or, just C, right now..), but that should be good enough
i know we talked about figuring out what calls we made in hydra/islandora/etc and using that as a basis
but i might be too lazy for that17:17
<barmintor>if you establish a baseline with ingest, we can elaborate to other operations later
that would be the most honest appraisal
<cbeer>cool. that sounds more fun than fighting with python and databank
<barmintor>it's easy_install all the way down17:18
<cbeer>(although i have learned that pip does know how to go fetch stuff from a git repo)
<barmintor>eddie was talking about trying to put together a homebrew package for it, once we figure it out17:19
<cbeer>maybe. i had a hell of a time with the homebrew python stuff though
<barmintor>since it will be hard to get people to hack on something that's this much of a pain to instal
<cbeer>but maybe i was thinking too far ahead
i wanted to use virtualenv
because i know all the cool kids use it
<barmintor>is it like RVM for python?
<barmintor>I'm hardly paying attention to what things do, just what I need to do to install/run them
<-- worst sysadmin ever
(probably not the worst)17:21
<cbeer>ok, i'll give databank another go now that i know more about pip and then i'll go back to jmeter
maybe i'll discover a better jmeter client than the awful swing one.17:22
<barmintor>there's probably an eclipse plugin somewhere
<cbeer>that's what i was thinking.17:23
ok, that's more promising:17:25
ImportError: No module named rdfdatabank.config.environment
ah, i guess it's not set up for running stand-alone, huh?17:28
i have to put it behind mod_wsgi?
<barmintor>yeah, that's next on my list
really just need to figure out what directory to mount
<cbeer>and, ew... i hope hard-coded paths don't go beyond those distributed wsgi files17:29
ok, want me to leave that to you, or should i give up and pretend to be a java developer?
(i feel like i stand 1/2 a chance of figuring out the wsgiref stuff)
<barmintor>I'm still working through the script I put together this weekend, so feel free17:30
to be brutal, i'm not liking databank. maybe if we truly though we could black-box it, it could be ok17:31
but if we're this deep in the guts just to make it go...
maybe we just need benosteen around to tell us what we're doing wrong17:32
i will say, repoze is pretty cool. at least i feel like i understand it better than spring17:33
<barmintor>yes, I think the benosteen idea is true17:35
have to say, a pure ruby app with a pocket solr/fedora seems pretty low maintenance here17:36
though I guess if I was setting up passenger to do development Iwouldn't feel that way
Hmm… solr schema, no solroconfig17:55
where did homebrew put solr?
<cbeer>barmintor: find it?18:48