Weekly check in 2011.12.15

• 13:31 <demory> Well it's about that time. Hello everyone!
• 13:31 <novalis_dt> Hi!
• 13:31 <Peejay_> hi!
• 13:32 <demory> Anyone else we know we're expecting today?
• 13:33 <novalis_dt> I assume Andrew will be by but I haven't heard anything
• 13:33 <demory> yeah, me neither. why don't we go ahead and get started
• 13:33 <demory> My check-in: mainly focused on the automated deployment work -- the graph builder automation is complete, now just need to plug that into the larger messaging-based framework in the OTPSetup repository. Also beginning to work on some front-end tools for the visualization/analysys api additions David has been checking in recently. Nothing directly TriMet-related to report this week.
• 13:34 <novalis_dt> I've been trying to get Andrew's MultiObjective pathfinding to do walking transfers on bike-transit trips, but I've been having some troubles
• 13:35 <novalis_dt> Along the way, I discovered a class of trips which it won't plan.
• 13:35 <novalis_dt> I've also been working on analysis tools
• 13:35 -!- grant_h [d819d25c@gateway/web/freenode/ip.216.25.210.92] has joined #opentripplanner
• 13:35 <novalis_dt> Frank, if you remind me of your analysis wishlist, I'll see if I can get any of that done
• 13:36 <novalis_dt> Something to do with transfers.txt iirc?
• 13:36 <novalis_dt> Anyway, that's me
• 13:36 <demory> novalis_dt will http://www.opentripplanner.org/apidoc/ update automatically to reflect the new analysis calls?
• 13:37 <novalis_dt> Nope
• 13:37 <novalis_dt> But you can generate your own docs
• 13:37 <novalis_dt> via mvn enunciate:enunciate
• 13:37 <demory> ok i'll look into that
• 13:37 <novalis_dt> Andrew updates it manually when we cut a release
• 13:38 <demory> maybe it would be helpful to have separate "development" api docs if we're going to be updating it more frequently now
• 13:38 <demory> i can take ownership of that if you think it's a good idea
• 13:38 <novalis_dt> demory, maybe, but I would hate to see people writing code against an API only for me to pull the rug out from under them
• 13:39 <demory> well, yeah, that's true
• 13:39 <demory> i suppose someone
• 13:39 <novalis_dt> I don't want to declare an API even a little bit usable until we know that it's really usable.
• 13:39 <demory> ..could always just look at the code
• 13:39 <demory> if they're really interested
• 13:40 <demory> Ok I'll hold off on that
• 13:40 <novalis_dt> If you have a problem with enunciate, let me know. It can be finicky.
• 13:40 <demory> ok
• 13:40 <novalis_dt> But their mailing list is really, really useful
• 13:40 -!- FrankP_ [1815504e@gateway/web/freenode/ip.24.21.80.78] has joined #opentripplanner
• 13:41 <novalis_dt> FrankP, I saw you were making some changes... Anything exciting?
• 13:42 <novalis_dt> Or, grant, mele, peejay, any bugs to report?
• 13:42 <FrankP_> Well, Printing is working now...plus the :: geo / name stuff. We released a new version of rtp.trimet.org the other day.
• 13:43 <novalis_dt> Yay, printing!
• 13:43 <novalis_dt> How's the new release going?
• 13:43 -!- mele [d819d1e5@gateway/web/freenode/ip.216.25.209.229] has quit [Ping timeout: 258 seconds]
• 13:44 <Peejay_> mele tends to know the bugs but she just ran away
• 13:44 <grant_h> no real bugs to report right now
• 13:44 <novalis_dt> Well, that's reassuring!
• 13:44 <Peejay_> but yeah, i think it's pretty good
• 13:44 <grant_h> she's here, her computer crashed
• 13:44 <grant_h> she says nothing to report either
• 13:45 <demory> just checked out the printing, look great!
• 13:46 <FrankP_> It's good...we get ~100 rtp.trimet.org users a day. Come next April, we should be ready to replace maps.trimet.org (with ~2000+ users per day, so we'll really see than). Next two tasks: our own OSM tiles, and our own geocoder (as opposed to ATIS geocoder)
• 13:46 <novalis_dt> I'm very excited to see the geocoder work
• 13:46 <novalis_dt> That's an area that people ask about a lot.
• 13:47 <FrankP_> thanks demory...still want to do a bit more with printing ... Valencia Spain has each leg with it's own little map...
• 13:47 <FrankP_> Printing is a big thing that's asked for...
• 13:48 <FrankP_> Problem point folks always seem to have is the geocoder. And mobile apps will always be needed...
• 13:49 -!- mele [d819d1e5@gateway/web/freenode/ip.216.25.209.229] has joined #opentripplanner
• 13:49 <novalis_dt> Yep, that's another area we would like to do more on
• 13:50 <novalis_dt> OK, does anyone have anything else?
• 13:51 <demory> oh, one quick thing
• 13:51 <demory> i've been looking into the steep streets issue
• 13:52 <novalis_dt> Any results?
• 13:52 <demory> i think I've identifies on strategy to at least cut down on those, which is to decrease the elevation sampling frequency along each edge
• 13:52 <novalis_dt> To 20m?
• 13:52 <demory> so instead of sampling every 10m, say, do 25m instead
• 13:53 <demory> it cuts down the # of warnings quite a bit, but doesn't completely eliminate them
• 13:53 <demory> so i still have some other work to do -- I'll send out a more detailed writeup in the next week or so
• 13:53 <novalis_dt> What about increasing sampling frequency, but using a low-pass filter or something?
• 13:54 <mele> 25 m is a long way though. I can think of some blocks/climbs that are significant that are shorter than that
• 13:54 <demory> yeah, i know. on very short edges you would still be sampling at least the start and end
• 13:55 <demory> also, I want to start experimenting w/ 1/9-arcsecond data. it is available for pdx and many other core urban areas
• 13:55 <demory> that could make a big difference too
• 13:56 <demory> but i'm pretty sure this is a limitation of the data as opposed to an algorithmic issue
• 13:56 <novalis_dt> demory, yeah, that's not crazy. I had considered trying to use it where available, but I didn't have a good strategy.
• 13:57 <demory> yeah we need to think that through some more for the general case. but at least for testing purposes i can acquire the pdx data manually
• 13:58 <demory> anyway look for a follow up on that from me
• 13:58 <demory> so, anything else?
• 13:58 <novalis_dt> Nope
• 13:59 <FrankP_> Nothing else from me.
• 13:59 <demory> ok well normally we'd have a call next Thurs (12/22) -- will people be around?
• 13:59 <demory> i will
• 13:59 <mele> we will
• 14:00 <FrankP_> yup, I'm here...kinda prefer chat though, to tell the truth.
• 14:00 <novalis_dt> I don't know if I'll be a round
• 14:00 <novalis_dt> Maybe.
• 14:01 <demory> FrankP so you'd prefer we chat every week?
• 14:01 <Peejay_> it's certainly easier
• 14:01 <demory> i guess the alternating arrangement was originally intended as a "trial" thing
• 14:02 <mele> there isn't as much urgent TriMet stuff going on right now, so it doesn't seem necessary at the moment
• 14:02 <mele> i imagine we could want to ramp it up again before the full release
• 14:02 <demory> yeah, the phone calls were more helpful during the main launch buildup
• 14:02 <FrankP_> Yes...although maybe a phone call every once in awhile would be a good thing.
• 14:03 <mele> monthly or something?
• 14:03 <demory> ok well I'm fine w/ going to chat every week (w/ occaisional calls as needed) -- anyone have an objection
• 14:03 <demory> ?
• 14:03 <novalis_dt> no objection
• 14:03 <FrankP_> nope
• 14:03 <Peejay_> no objection either!
• 14:04 <demory> maybe just play the call thing by ear -- we'll probably want to time them w/ major milestones, etc
• 14:04 <FrankP_> sounds good, david
• 14:04 <demory> but on average monthly sounds reasonable to me
• 14:04 <demory> ok great
• 14:04 <demory> so we'll be back on IRC next Thurs
• 14:05 <demory> have a good week!
• 14:05 <novalis_dt> Later!
• 14:05 <Peejay_> bye
• 14:05 <FrankP_> take care, all