2011.07.28 Weekly Check In - mattwigway/OpenTripPlanner GitHub Wiki
13:32 <demory> hello all
13:32 <novalis_> hi demory
13:33 <demory> shall we get started?
13:33 <FrankP> hi David
13:33 <novalis_> Sure!
13:33 <demory> ok
13:33 <demory> anyone else there?
13:34 <demory> i guess it's just us for now
13:34 <demory> maybe we can start with quick check-ins
13:34 <demory> i'll start
13:34 <demory> i've mostly been working on admin / non-technical stuff
13:34 <demory> the website, roadmap, etc.
13:35 <demory> but i'm beginning to work on the Euro Challenge demo
13:35 <demory> pulling the base data out of OSM as we speak
13:36 <demory> so hopefully we'll know if that's going to be feasible within a couple days
13:36 <novalis_> demory, you might want to talk to FrankP about that, since I think he has a better process for that
13:36 <novalis_> (the base layer)
13:36 <novalis_> I've been working on miscellaneous routing bugs and suboptimalities.
13:37 <demory> ok let's come back to that later
13:37 <novalis_> In particular, I've been working with the TriMet interns on issues around narratives and bike routing
13:37 <novalis_> I've also been chatting with Andrew about the dread #409
13:37 <novalis_> He has been having some git/maven/eclipse issues
13:37 <novalis_> Has anyone else had issues with eclipse and/or maven since the move to git?
13:38 <FrankP> only 1 day back from vacation since the workshop -- this week is moving current map & otp over to new production servers. Need to get git repo of OTP in place of existing .svn stuff (e.g., cron to pull latest code, and rebuild code and graph).
13:38 <demory> honestly I haven't really been working with the code much
13:38 <demory> but i'll let you know if i run into anything
13:40 <demory> ok thanks for the updates
13:40 <demory> can we talk about the base layer for the Pacific NW demo?
13:40 <FrankP> as to building a map base layer from OSM, I'm not doing that...I still use the Portland Metro data.
13:40 <demory> my plan was to define a polygon that is a buffer of all the routes we're looking
13:41 <demory> and then use osmosis to extract from planet.osm
13:41 <novalis_> FrankP, But you're using OSM for routing, right?
13:42 <FrankP> yup. big bounding box pull directly from their server nightly (bbox includes Clark County in WA, and down south past Salem ... way beyond our bus routes).
13:42 <FrankP> not planet.osm...directly from a rest server...I'll get the URL.
13:43 <demory> ok
13:43 <demory> yeah if I can avoid working with planet.osm that would be nice
13:50 <kpw> might be other similar sources too but this has worked for my purposes
13:50 <novalis_> We really ought to have something on the wiki about this
13:50 <demory> 10 sq deg may be too small, i'll have to look into that. but it sounds like merging the state files could work too
13:51 <demory> in any case i still want to extract the buffered area at the end. i want to keep this graph as small as possible
13:51 <novalis_> Speaking of which, I think there is some other wiki cleanup that could be done -- in particular, adjusting the checkout instructions to no longer mention subversion. Does someone want to take the lead on that?
13:51 <demory> i can do that
13:52 <demory> i've been doing some cleanup on the wiki but there is much more to do
13:52 <demory> a lot of the formatting -- spacing, etc. -- needs to be tweaked
13:52 <FrankP> and the front-page image is too big :-0
13:52 <novalis_> I actually thought of a way to make the graph a tiny bit smaller, if you end up just a bit too tight: simply replace every instance of CAR in the DefaultWayPropertySet with NONE.
13:53 <novalis_> Yeah, the spacing got a little wacky during the import
13:53 <demory> and some of the stuff is out of date (old roadmaps, etc)
13:54 <demory> i will fix the image too :)
13:54 <kpw> i think the pdm merge/extract step could also be built into the graph builder (maybe).
13:54 <novalis_> Yeah, Trac was shrinking that image down to 400px wide, but github didn't have that feature.
13:55 <novalis_> kpw, there was a ticket on that, but we ran into some trouble getting the protocol buffers stuff mavenized. Since then, the situation seems to have improved, so maybe it's worth reexamining that ticket
13:55 <kpw> at the very least, i'll document my steps for the dc demo to explain how to do it manually
13:55 <kpw> cool
13:55 <novalis_> I would feel very good about having that old ticket finally killed.
13:56 <FrankP> did someone say, "protocol buffers"?
13:56 <kpw> yes re OSM
13:56 <demory> i agree but i still want to do this manually w/ the PNW demo. i need practice w/ this stuff
13:56 <kpw> the PBM files are proto buffs
13:56 <kpw> yeah, it's not too tricky (at least with the bbox extract) and is worth trying out
13:58 <demory> ok well i have some things to try with that. anything else on the Euro Challege?
13:58 <demory> did people see the latest edits to the letter by Wojciech?
13:58 <demory> i think it looks good so far
13:59 * novalis_ looks
14:00 <demory> well we have some time on that. i'm more concerned about the demo right now
14:00 <demory> i'll keep you all posted on how that goes
14:00 <novalis_> Sure. We should also think about figuring out how we are going to manage configurations and builds on demos.
14:11 <demory> we were just talking about tickets. they're now visible on GitHub if you havent seen them yet
14:12 <andrewbyrd> novalis_: thanks, I was about to ask for that
14:12 <novalis_> yeah, Andrew and I had spoken about the issue situation
14:12 <novalis_> So, now how do we turn off issues in trac?
14:12 <novalis_> Also we should remove some of the really old ones that aren't relevant to the modern world.
14:13 <novalis_> ie #8
14:13 <demory> so we're completely done with them in trac? all of the discussions, etc. have been imported?
14:13 <novalis_> Looks like it!
14:14 <andrewbyrd> I moved some of the longer tickets over to wiki pages
14:14 <demory> ok great
14:14 <andrewbyrd> there are a probably a few more that could potentially be in that format, but most of the current debates that didn't belong in tickets are in wiki pages
14:16 <demory> yeah i see a few that have 10+ comments. novalis_ now i see what you mean about you owning all of them
14:17 -!- kpw [46157fcc@gateway/web/freenode/ip.70.21.127.204] has quit [Ping timeout: 252 seconds]
14:17 <andrewbyrd> I was thinking the import would be more tolerable if it was not david but some generic user who created all the tickets, so it didn't give the impression of one user posting everything
14:18 <andrewbyrd> but once the issues are created, there is no means to remove them through the API
14:19 <novalis_> I can live with being a bit too loquacious
14:21 <demory> yeah i think we can work around that. going forward others will start creating their own and it will look more like a multi-person effort
14:21 <novalis_> And hopefully, we'll close some of those old tickets!
14:21 <demory> also the tickets can still be assigned to anyone regardless of who owns them right?
14:22 <novalis_> Right.
14:22 <demory> yeah i will start going through them looking for candidates for closure. (yes #8 definitely qualifies)
14:22 <andrewbyrd> yes, I think the only disadvantage is that you have to check the bottom of the comments to see who wrote them
14:22 <novalis_> Actually, FrankP, I don't see you on the OTP team on github -- can you make an account?
14:23 <novalis_> Also, there seems to be a bit of a problem with the import where some closed tickets didn't really get closed.
14:23 <demory> hmm
14:23 <novalis_> but that's fixable
14:23 <demory> i guess we should keep trac online a little bit longer
14:24 <demory> so I can cross-reference between it and github
14:24 <novalis_> Sure.
14:26 <demory> that reminds me, i've been talking to Evan about the new website. i think he's going to create a dev instance of wordpress for me to work with
14:26 <novalis_> ok, cool.
14:26 <demory> so nothing will change at otp.org right away
14:27 <demory> and even when we do launch the new site, we might be able to keep trac running in the background if needed
14:27 <demory> but it sounds like we may not have to
14:27 <FrankP> novalis_, I'm on github and 'watching' OTP...don't see a OTP team list nor how to join
14:28 <novalis_> Ok, maybe I need to add you or something
14:28 <FrankP> account is fpurcell
14:28 <novalis_> demory, yeah, the tickets are the last thing we care about trac for
14:28 <novalis_> Oh, I just hadn't added you. Sorry about that!
14:29 -!- vdeparday [[email protected]] has quit [Remote host closed the connection]
14:29 <novalis_> I just assigned you an issue -- does that show up for you somewhere?
14:30 <novalis_> Also, what else is on the agenda for today's meeting?
14:31 <demory> i really don't have anything else
14:31 <andrewbyrd> Going back to the maven/git issues you mentioned earlier
14:32 <novalis_> Yeah...
14:32 <andrewbyrd> I have run into quite a bit of flakiness in different forms
14:32 <novalis_> In my email, I actually forgot to mention that I always run mvn test from the command line, although I run individual tests from Eclipse's JUnit runner
14:32 <andrewbyrd> I get the impression they overhauled a few things in the new Elcipse release, possibly to fix some of this, but adapting to that version is going to involve redoing some details of the POM
14:33 <novalis_> That doesn't bother me in principle -- only in practice, I am not great at Maven
14:33 <novalis_> Do you think bdferris would be able to do it?
14:34 <andrewbyrd> I've got it mostly working, the remaining problem is that when I do a build from the command line everything is fine, but nothing happens from inside eclipse
14:34 <andrewbyrd> not nothing, there is acivity on the maven console, but it is not a full build
14:34 <novalis_> When you look at that activity, does it tell you what's wrong?
14:35 <andrewbyrd> No errors.
14:35 <andrewbyrd> I believe it's simply not building the submodules
14:36 <novalis_> I had so many problems with Helios that I'm reluctant to try Indigo
14:37 <andrewbyrd> Anyway, apparently there's not much to compare with, and I suppose I need to figure this out on my own
14:37 <novalis_> But I guess it won't hurt to download it just for testing...
14:37 <andrewbyrd> The problem is it's getting in the way of me testing my changes
14:37 <novalis_> Can you test manually on Tomcat?
14:38 <andrewbyrd> But I'm not even on Indigo. I tried that once, and switched back immediately once I realized I was going to need to rewrite the poms
14:38 <andrewbyrd> So I'm on Galileo
14:38 <novalis_> Oh!
14:38 <novalis_> So the only change is the change to git?
14:38 <andrewbyrd> Using the ubuntu package that used to work find
14:38 <andrewbyrd> Using the ubuntu package that used to work findfine
14:38 <novalis_> But that doesn't make any sense -- it's the same source code...
14:39 <novalis_> Did you try creating a new workspace and importing the projects (as maven projects) into that workspace?
14:40 <andrewbyrd> I think I tried every possible combination of importing maven, git and basic projects, adding git control or maven dep management after the fact, with and without the submodules as top-level projects... usually doing the git clone on the command line but also from the git repo view...
14:40 <novalis_> Hm!
14:41 <novalis_> Ok, did you try all that with the second-to-latest SVN version to prove that it's the switch to git that's the problem?
14:41 <andrewbyrd> Anyway, I don't expect you guys to come up with a solution at this second, since you haven't been running into this. I wonder if there are remnants of my Indigo install.
14:41 <novalis_> Oh, that could be.
14:42 <andrewbyrd> Except that it was a download from the eclipse site, in its own little directory (not a package), but that doesn't mean it didn't decide to leave who knows what elsewhere...
14:42 <andrewbyrd> Still, it's doing this to me on two different machines
14:43 <andrewbyrd> You're right, clearlythe thing to do is to try again on svn and see if it works correctly
14:44 <andrewbyrd> And yes, I will just do the maven and deployment stuff on the command line for the time being
14:44 <andrewbyrd> I don't really think this is a problem directly related to git, it's a quirk somewhere that just needs to be identified and avoided.
14:45 <novalis_> Right.
14:45 <andrewbyrd> One other thing: the docs on defaultWayPropertySetSource
14:46 <novalis_> Are they not correct?
14:48 <andrewbyrd> I think they are fine, but there are xml samples on the wiki page that still include defaultAccessPermissionsSource
14:49 <novalis_> Oops!
14:49 <novalis_> Will fix
14:49 <andrewbyrd> Now that I do a search, it's a small change, I should have just fixed it instead of complaining
14:50 <novalis_> I already have the wiki checked out.
14:50 <andrewbyrd> but I just remembered noticing that
14:50 <novalis_> Looks like just one, right?
14:51 <novalis_> (and I just fixed that one)
14:52 <novalis_> I love that github wikis are in git so I can just go in with Emacs and fix stuff
14:52 <andrewbyrd> the "Permissions and bicycle safety" mentions defaultAccessPermission
14:52 <novalis_> Oh, one more thing that didn't get imported to github issues from trac: milestones
14:53 <novalis_> andrewbyrd, I don't see that.
14:54 <andrewbyrd> Would it make more sense to redo the milestones, rethinking them in our current context rather than importing them?
14:55 <andrewbyrd> novalis_: yeah, it's no longer there. looks good.
14:56 <FrankP> novalis_, issues view lists the one issue assigned to me....that said, I'm unable to fix this particular issue re right-click, so can you assign it to someone else and/or close it.
14:56 <novalis_> Unable because you don't have time, or because it's too technically difficult, or because it's a bad idea?
14:57 <FrankP> all three
14:58 <novalis_> Can you make a comment to that effect and then close it?
14:58 <FrankP> sure.
14:58 <novalis_> Thanks.
14:59 <demory> re milestones, I can take a look at that. per Andrew's suggestion, this could be a good chance to reassess that
15:00 <demory> so anything else while we're all here?
15:00 <andrewbyrd> There are no links into this page yet because I wanted to clean it up a bit, but if we are now using github issues I'll link it in.
15:01 <demory> sounds good, thanks
15:01 <novalis_> I have created two milestones: TriMet demo, and Future. Do we need more milestones than that?
15:02 <demory> that's fine for now. Trimet demo is obviously the immediate focus
15:02 <andrewbyrd> demory: I guess our plan from last week about copying tickets manually is no longer relevant, so you no longer need me to give you a list of the tickets that were wikified
15:02 <demory> right
15:03 <novalis_> We should make sure they are closed in the github tracker
15:03 <andrewbyrd> Agreed, I think that's enough milestones for now
15:03 <andrewbyrd> right, I closed them as I was going along in trac but you imported to github before that
15:03 <demory> once we have a clearer picture of future plans re analytical tools, system mapping, etc. we may want to break the future one up into mutiple
15:03 <andrewbyrd> I'll take care of closing them, I have a list
15:04 <novalis_> demory, that's reasonable.
15:04 <demory> but that will come later
15:05 <novalis_> Maybe we need one more milestone for "on fire"
15:05 <novalis_> i.e. #409
15:06 <andrewbyrd> That also sounds reasonable
15:06 <demory> do github issues have a priority/urgency field like on svn?
15:06 <andrewbyrd> for 409 itself, the code changes are very simple, but I got mired in this maven build mess
15:06 <novalis_> I don't see that...
15:08 <demory> there does seem to be a concept of issue "labels" -- so we could track especially urgent ones that way too
15:08 <demory> i.e. perhaps "on fire" should be a label not a milestone?
15:09 <novalis_> I actually think the key questions an issue tracker answers are, "what should I work on next?", and "when will this bug be addressed?"
15:09 <andrewbyrd> I see that there is now an on fire milestone :)
15:09 <novalis_> So having the order of work be "on fire", "trimet demo", "future" makes a lot of sense to me
15:10 <novalis_> For labels, we should probably restore the old "which component is this in" labels
15:10 <andrewbyrd> to me labels are more for specifying which components are concerned, so people can spot things they understand best
15:10 <demory> ok that sounds reasonable
15:10 <FrankP> trimet beta (instead of demo
15:10 <novalis_> Fixed
15:13 <andrewbyrd> So, just to confirm: we are doing everything on github now. trac is purely for reference?
15:13 <demory> that is my understanding, yes
15:13 <novalis_> Yep.
15:13 <demory> also i can work on assigning labels to components as i go though the tickets looking for ones to close
15:15 <novalis_> Cool!
15:15 <FrankP> One last one, demory. Anyone hear from Google about GTRTFS? (Bibi said a few weeks back we're waiting on Google to make their spec public) My #1 ticket is https://github.com/openplans/OpenTripPlanner/issues/412, and I hope we can get that into the TriMet beta.
15:17 <novalis_> demory, I can't remember if you are familiar with the issue -- if not, I can fill you in.
15:17 <demory> I have not heard anything from Google. But I'll look into to it
15:18 <demory> i remember it being discussed at the conference but haven't heard anything since
15:18 <novalis_> Thanks.
15:18 <demory> let me check w/ Kevin, perhaps he has more recent info
15:19 <demory> but i have assigned the ticket to the Trimet beta
15:21 <demory> ok, i say we call the official check-in over. i will post transcript to wiki
15:21 <novalis_> Cool.
15:22 <andrewbyrd> ok thanks
15:22 <demory> i should be on irc for a while longer though