The IRC channel is #sakai on the freenode network, you can join using a web client at http://webchat.freenode.net/?channels=sakai and you can view the complete logs at http://sakai.iriscouch.com/irc/_design/viewer/index.html
08:49 <SakaiGithub> [3akai-ux] bp323 pushed 2 new commits to master: https://github.com/sakaiproject/3akai-ux/compare/f83d9dea2222...40a2620f1e2f 08:56 <SakaiGithub> [3akai-ux] bp323 pushed 1 new commit to master: https://github.com/sakaiproject/3akai-ux/commit/6befb0f4d2f37a6de4b57c3f99fd8e8fc58a7ee4 08:56 <SakaiGithub> [3akai-ux/master] SAKIII-5843 added screen reader text where a dropdown arrow is - jsloane 11:07 <mrvisser1> PhysX: Thanks for looking at the PR for me 11:08 <PhysX> mrvisser1: sure, I don't have commit access to that repository for some reason 11:08 <PhysX> mrvisser1: I asked Nico give me some 11:09 <mrvisser1> PhysX: saw that, thanks. have you used "optimist" before? nice module for handling command-line arguments. 11:09 <PhysX> mrvisser1: Never used it before but it looks promising 11:09 <mrvisser1> PhysX: in case you haven't seen the PR, I changed generate.js and loaddata.js to use it. 15:23 <simong> Do we expose the solr endpoint that a default solr install exposes when using the embedded solr option? (ie: http://localhost:8983/solr) 15:28 <zacht> simong I'm pretty sure the answer is no, but I don't know whether there's any technical reason for that or not. 15:29 <simong> mmh, that's what I thought 15:29 <zacht> simong i.e. I don't see why it wouldn't be possible. 15:30 <simong> Any idea of how hard it would be or where I should start? 15:31 <zacht> simong Yes, you should start by asking thecarlhall or raydavis. :-) 15:32 <zacht> simong Or _mst, but you two are in fairly incompatible time zones… 15:32 <simong> ok 15:32 <simong> I'll try to set up a remote one, as that is how it should be run in production anyway 15:32 <simong> (for now) 15:33 <zacht> simong thecarlhall said he's working on some automation to make the standalone server configuration easier for everyone. 15:33 <simong> cool, that'd be useful 15:34 <zacht> simong It helps that we just moved up to the current solr release. 15:41 <raydavis> simong, definitely a good thing to use the separate Solr server if you're going to expose the 8983 port access. For one thing, you get access to Solr's admin console. :) 15:42 <simong> ok :) 15:55 <RyanEduSys> I'm having trouble with sakai hybrid. I installed it and everything seemed to work perfectly, but now trying to start it up again, it's failing to start CLE. Here's my output: http://pastebin.com/brAChnvU 16:46 <zacht> dgillman ctweney simong Check this out: https://jira.sakaiproject.org/secure/RapidBoard.jspa?rapidView=1&quickFilter=6 16:47 <zacht> I've been looking at GreenHopper. 16:47 <zacht> This is like our Trello board: you can drag your issues between columns. 16:47 <zacht> (though it's a little harder to look at) 16:49 <zacht> I've created a version for nakamura called "Sprint 150", which corresponds to any work to be completed before the standup meeting on Thursday. 16:49 <simong> I find it a bit clearer to look at than trello. 16:49 <zacht> simong Oh, good! :-) 16:50 <simong> It doesn't hide it's data in miniature windows so you can actually scroll the page ^^ 16:50 <simong> I only see on task that is still left to do though, is that correct? 16:51 <simong> s/on/one 16:51 <zacht> simong No way! 16:51 <simong> and 6 that are under review 16:51 <zacht> simong The Sprint 150 version for KERN is only a few minutes old. 16:51 <simong> ah oops ok ^^ 16:52 <dgillman> zacht: that's cool :) 16:52 <zacht> simong I'm pointing this out now so that any work you're doing now that stands a chance of being wrapped up by late Thursday, you set its fix version to "Sprint 150" 16:53 <simong> ah gotcha 16:53 <zacht> simong dgillman ctweney There's also a "Sprint 151" which will correspond to the two weeks following Thursday. 17:05 <ctweney> zacht: ugly as it is that Jira rapid board thing has the advantage of being tied in to our existing data and workflow 17:05 <ctweney> tho I have not tried to actually use it yet 17:05 <ctweney> just glanced at it 17:06 <zacht> ctweney Yes, it seems to map nicely to the way we're working already. 17:06 <ctweney> why can't I drag/drop tho? 17:07 <zacht> ctweney I thought we were supposed to be able to. 17:07 <ctweney> I can drag within a category eg "Everything Else" 17:07 <ctweney> but can't move from "Everything Else" to "Expedite" 17:07 <zacht> ctweney Ah. Expedite is computed automatically. 17:07 <ctweney> aha 17:08 <ctweney> based on…? 17:08 <zacht> ctweney From its priority. 17:08 <ctweney> ah so 17:08 <ctweney> makes sense 17:08 <zacht> ctweney They call those "swim lanes" 17:08 <zacht> ctweney and they're configurable 17:08 <zacht> ctweney Expedite is just the one that's in there by default. 17:09 <zacht> ctweney And it's currently set up to trigger on anything that's Blocker. 17:10 <zacht> ctweney We could have a swim lane with just the Sprint 150 issues for example. 17:10 <zacht> ctweney I haven't fully grokked the best way to use it yet. 17:11 <zacht> mrvisser We're talking about Jira's "rapid board" https://jira.sakaiproject.org/secure/RapidBoard.jspa?rapidView=1&quickFilter=6 17:12 <zacht> mrvisser I've created two new versions for KERN: Sprint 150 and Sprint 151, so please use those where appropriate. 17:37 <dgillman> zacht: I'm on hold in both my current tasks; thoughts on something I might pick up before the close of this sprint? 17:38 <zacht> dgillman How about a well-deserved break from the search bundle? 17:38 <dgillman> zacht: :) 17:39 <dgillman> I was eyeing the Document/test current pooled-content functions backlog item 17:39 <dgillman> but that was just because it looked like something on which progress could be made in the short time before the close of sprint 17:40 <zacht> dgillman You should coordinate with ctweney, since he's working in that area. I'm sure there is divisible work. 17:40 <dgillman> zacht: sounds good 17:41 <dgillman> ctweney: you around, bud? 17:43 <ctweney> I'm here dgillman 17:44 <ctweney> what task or jira specifically were you looking at? 17:44 <dgillman> ctweney: I was looking at the Trello board backlog 17:44 <dgillman> ctweney: "Document and/or test all the Pooled Content operations we currently support." 17:44 <ctweney> oh, that's way out of date 17:45 <dgillman> joy! 17:45 <ctweney> I have not bothered updating Trello to match what I'm doing in jira 17:45 <ctweney> that work is kern-3070 and subtasks which are mostly done 17:46 <dgillman> well, I was looking for something to carve of before the close of sprint. any thoughts around pooled-content tasks? 17:47 <ctweney> well dgillman my next step is to extract a service interface for pooled content … which will take me all through files/impl and is pretty much a one-man job 17:47 <ctweney> other work on files/impl might get in the way of those refactorings 17:48 <dgillman> ctweney: makes sense 17:48 <zacht> dgillman ctweney I think we'll deprecate the Trello board within a day or so. 17:48 <ctweney> however 17:48 <ctweney> it looks like "build tests to generate pooled content load" might make sense 17:48 <ctweney> if mrvisser's test environment is far enough along that is 17:49 <zacht> ctweney dgillman You can probably make good progress on that without necessarily having the environment set to go. 17:49 <dgillman> ctweney: is should be possible to start constructing the tests even if he's not ready 17:49 <dgillman> ctweney: yeah, what zacht said 17:49 <dgillman> :) 17:50 <ctweney> agreed 17:50 <dgillman> ctweney: is there a Jira for that already? 17:50 <zacht> ctweney dgillman I think the hard part is capturing the requests in the right sequence which constitute realistic uses of pooled content. 17:50 <ctweney> not one that I know of. 17:50 <zacht> dgillman The cucumber tests that ctweney created will probably help out a lot. 17:51 <ctweney> well, be careful there, dgillman, I wrote those tests to exercise the whole bundle, not necessarily to be a realistic set of usages 17:51 <ctweney> however, there are some building block cuke features that might be reusable 17:52 <dgillman> zacht, ctweney: got it. so the focus should be building cuke tests that reflect "realistic usage" for load testing of files… however we might decide to gloss "realistic usage" 17:52 <ctweney> yeah 17:52 <zacht> dgillman I wouldn't use cucumber for that task. 17:52 <zacht> dgillman try tsung. 17:52 <dgillman> zacht: tsung, got it 17:53 <zacht> dgillman The cool thing is, the cucumber step definitions might actually be usable from within the code that drives tsung 17:54 <dgillman> zacht: where are load testing scripts being stored? I see testscripts/jmeter…. where are existing cuke scripts? 17:55 <zacht> dgillman testscripts/integration 17:55 <dgillman> coolio 18:00 <zacht> dgillman You can see them in action like this: cd testscripts/integration; bundle install; bundle exec cucumber 18:01 <dgillman> zacht: sounds good. Am just skimming Tsung docs. So I understand, why are you suggesting Tsung over Cuke? 18:02 <zacht> dgillman cucumber doesn't simulate a load. 18:02 <dgillman> zacht: :) that would do it 18:05 <zacht> dgillman Here's a pair of screencasts I made a while back about tsung: http://screencast.com/t/2jJOu3DDxtx 18:06 <zacht> dgillman part 2: http://screencast.com/t/mp9Ux2Y5vD 18:06 <dgillman> nice! zacht++ 18:07 <GitHub127> [nakamura] zathomas pushed 4 new commits to master: https://github.com/sakaiproject/nakamura/compare/0be8a5af19a5...4fc08a4abdc2 18:07 <GitHub127> [nakamura/master] KERN-3074 Remove immediate=true where not needed. - Carl Hall 18:07 <GitHub127> [nakamura/master] KERN-3076 Remove metaype=true if no configurable properties or just system properties. - Carl Hall 18:07 <GitHub127> [nakamura/master] KERN-3074 Have ServletDocumentationRegistryImpl activate immediately to save circular thread check exception on delayed services in Felix. - Carl Hall 18:09 <zacht> dgillman And there's Kyle's work, which encapsulates tsung tests with ruby: https://github.com/kcampos/Open-Performance-Automation-Framework 18:28 <zacht> dgillman I just had an idea. Using ruby's Net::HTTP::Proxy, you could run all our integration tests (cucumber included) through tsung's recorder. 18:29 <zacht> dgillman Maybe that wouldn't be much like a production load, but it would be a lot of coverage at a low cost. 18:31 <dgillman> zacht: that sounds fine, but is a slightly different goal from developing realistic usage scripts for pooled-content. if these are separate goals I don't want to conflate them 18:32 <zacht> dgillman You're right. 18:32 <dgillman> zacht: rats, did I just decline an easier task ;) 18:32 <dgillman> heeh 18:32 <dgillman> hehe 18:35 <simongee> raydavis: can I pick your brain a bit about the external profile services? 18:36 <raydavis> simongee, sure, but there's not much there. I believe that's still unused code. 18:36 <simongee> ah ok 18:37 <simongee> I just wanted to confirm some reasoning 18:37 <GitHub199> [nakamura] zathomas pushed 1 new commit to master: https://github.com/sakaiproject/nakamura/commit/1445cb085ceb64d3baa3e23f9ecbe514b5d75e7e 18:37 <GitHub199> [nakamura/master] KERN-3070 resolve merge conflict and merge b2636db - Zach A. Thomas 18:37 <simongee> Is it correct that in order to retrieve data from an external source there needs to be a property on the authprofile (or one of it's children) ? 18:38 <simongee> I mean the sakai:source=external property 18:38 <raydavis> simongee, as I remember that was the theory. But since its usefulness was never verified by actual customers, I'm very skeptical. 18:39 <simongee> Yes, I agree 18:39 <GitHub116> [nakamura] thecarlhall pushed 2 new commits to master: https://github.com/sakaiproject/nakamura/compare/1445cb085ceb...eb7188b67f29 18:39 <GitHub116> [nakamura/master] KERN-3052 Make User ID searches more precise - Ray Davis 18:39 <GitHub116> [nakamura/master] Merge pull request #955 from raydavis/KERN-3052 - Carl Hall 18:39 <raydavis> In my experience, solutions without usage are broken. :) 18:39 <simongee> it seems backwards that an authprofile node expresses that we should grab data from elsewhere 18:39 <simongee> it does work (sort of) 18:41 <simongee> I'm able to retrieve data from VIVO (external app) if I write my ProfileProvider and place a sakai:source=external on every authprofile node and have a sakai:profile-provider=vivo on the /var/profile/providersByName node 18:41 <raydavis> simongee, I would assume that this whole area is up for eradication/rethinking as part of OAE's move to UX-tailored client-server data models -- might be a good thing to chat on-list about. 18:42 <raydavis> For us at Berkeley, we're more likely to need external group provision before external profile-section provision. 18:42 <simongee> sure, I'll post an email later tonight 19:09 <thecarlhall> that integration test we have for testing long strings in sparse does some serious flexing of osgi 19:10 <thecarlhall> it takes a while to restart almost every bundle in the server when modifying sparse like that 19:20 <zacht> thecarlhall Could we fix that with some careful bundle rearranging? In other words, keep that configuration from affecting a hundred downstream things? 19:21 <zacht> thecarlhall It's JDBCStorageClientPool, right? 19:21 <thecarlhall> zacht, splitting sparse into api/impl may help that but I haven't tested with the api/impl split 19:21 <thecarlhall> zacht, yeah, it's when the config on that class gets changed. The service doesn't have @Modified so it stops & restarts which twists up the whole OAE dep graph 19:22 <thecarlhall> zacht, maybe soon I'll be able to test some theories on the api/impl split. 19:22 <zacht> thecarlhall I think we'd see a number of benefits from that. 19:23 <thecarlhall> zacht, if it stops the graph from reloading, that'd be a huge win to me 22:21 <ctweney> hey thecarlhall, do you have a quick list of links that will help a total noob get up to speed on cassandra? 22:25 <thecarlhall> ctweney, hm, I'll have to dig around. 22:26 <thecarlhall> ctweney, I went through /a lot/ of presentations 22:26 <thecarlhall> ctweney, the twissandra example is a good starting place 22:26 <ctweney> datastax's site has some good articles 22:27 <thecarlhall> as your reading stuff, remember this: Super column family's are on the way out and composite columns are the new hotness 22:29 <thecarlhall> ctweney, http://www.slideshare.net/aaronmorton/cassandra-does-what-code-mania-2012 22:29 <thecarlhall> ctweney, http://www.slideshare.net/lomakin.andrey/apache-cassandra-part-2-data-model-example-machinery 22:55 <_mst> one thing I keep wondering with the twitter examples is how you'd design it to ensure that a tweet gets delivered to all the people it's meant to 22:56 <_mst> thinking about Stephen Fry tweeting to his 4 million followers, but the server handling his request getting restarted mid-tweet and ending up with only 2 million followers seeing the tweet 22:56 <_mst> maybe you'd have to roll your own transactions... write a sort of journal to a different CF and reapply it on startup or something? 23:05 <thecarlhall> _mst, with cassandra in that scenario, you would have 2+ servers and cassandra guarantees wrties 23:07 <_mst> but if the client doing the writes itself dies? Or would it send the 4 million tweet writes in a single request to cassandra? _______________________________________________ oae-dev mailing list [email protected] http://collab.sakaiproject.org/mailman/listinfo/oae-dev
