Weekly check in 2012.03.29 - GeoSmartCity-CIP/gsc-opentripplanner GitHub Wiki
13:33 <demory> hello everyone!
13:34 <grant_h> hello
13:34 <novalis_dt> andrewbyrd, what's up with Graph.nearestVertex?
13:35 <novalis_dt> Re meeting, I'm working on optionally treating transfers.txt as a source of pathways, to handle NYCT subways.
13:36 <demory> my update is pretty quick -- various workflow improvements and testing / debugging for OTPSetup. the public version at deployer.otp.org now supports fare factory selection, direct downloading of graphs. and a number of enhancements on the admin side
13:36 <demory> currently trying to figure out what's going on w/ a large (13 feed!) build we're trying for the NY/NJ area
13:37 <mele> sounds fun
13:38 <andrewbyrd> novalis_dt: it's a very inefficient method to find the nearest vertex. doesn't appear to be used anywhere. Why do you ask?
13:38 <novalis_dt> andrewbyrd, well, I thought you might have some use in mind for it -- if not, I'm going to delete it
13:38 <andrewbyrd> yeah, i would hate for someone to call that in a loop.
13:40 <andrewbyrd> I'm looking at some MOA* bugs, and looking into automated stop time filtering.
13:42 <andrewbyrd> we did another bugfix release yesterday
13:42 <andrewbyrd> and I just asked the OBA list about this string intern slowdown issue
13:43 <demory> re 0.5.3 -- you did turn string interning off, right?
13:43 <andrewbyrd> demory: yes, it's off.
13:44 <demory> ok, thanks
13:44 <andrewbyrd> and speaking of OBA it looks like we have an indirect dependency on a one-year-old version of the CSV reader... should we just move up to the most recent release of OBA-GTFS?
13:44 <demory> i don't think that's the problem w/ ny but I'll do some more testing after the current test build
13:45 <andrewbyrd> demory, I don't think it should cause any crashes. I've been doing NYC builds locally for a couple of days and varying that method.
13:45 <kpw> what's with the old dependency?
13:45 <novalis_dt> andrewbyrd, actually, I was going to upgrade OBA-GTFS anyway
13:45 <novalis_dt> Because the new version has a method I need.
13:47 <andrewbyrd> I don't know, but I just cloned OBA-GTFS and the csv reader to make a patch, and there has been a bit of change since the version we are using
13:50 <demory> OK any other check-ins / updates? TriMet folks, anything on your end?
13:50 <mele> we've been mostly focusing on making sure the license change doesn't break everything
13:51 -!- mattwigway [9912236f@gateway/web/freenode/ip.153.18.35.111] has joined #opentripplanner
13:51 <kpw> oh yeah, how big of an impact is that having?
13:51 <mele> everything looks good here now but it's a big issue in other places
13:51 <kpw> yikes!
13:51 <novalis_dt> Which license change?
13:51 <kpw> how much data is going away?
13:51 <mele> i'm not sure what the data situation is over there...
13:51 <mele> it depends, let me get a link for you
13:52 <grant_h> yeah, I know parts of California are going to lose a lot of data
13:52 <mattwigway> Is this the OSM license change we're discussing?
13:52 <mele> openstreetmap
13:52 <mele> they're deleting all edits by everyone that hasn't agreed to the new terms starting on sunday
13:52 <mele> and the database will be read-only for at least a few days
13:53 <mele> people have been working hard to get things caught up
13:53 <demory> perhaps this is a good time to refresh planet.osm on the deployer server
13:53 <mele> the "badmap" shows things that will probably go misisng
13:54 <mele> demory: yeah it would be a really good time to do that
13:55 <novalis_dt> Looks like we're losing some subway stations -- someone should run go-sync.
13:55 <mele> Los Angeles is in a world of hurt, and some other towns and cities, but mostly things look ok
13:55 <novalis_dt> Whoa, and Mosco St, where I get my thai groceries and 5/$1 dumplings!
13:55 <grant_h> I know South Carolina is in really bad shape as well
13:56 <kpw> that's really sad
13:56 <kpw> folks should have triaged this and tried to patch those gaps...
13:56 <kpw> this is going to be a big setback for osm
13:56 <mattwigway> Holy cow, whole streets missing in SF, and Golden Gate Park!
13:56 <novalis_dt> Are people just refusing to relicense? Or has nobody contacted them? Or do they juts not respond to email?
13:56 <mele> yeah i'm a little worried about how much it will piss off the data consumers
13:56 <kpw> it will big time
13:56 <mele> people have been trying to contact them for a long time
13:56 <kpw> google will be all over this
13:57 <mele> some have must moved on, some are refusing out of spite/silliness
13:57 <kpw> i think the osm folks should have pusehd to replace the important data
13:57 <mele> oh they have
13:57 <mele> there just aren't enough people to actually get everything done in time
13:57 <kpw> oh wow, yikes
13:58 <mele> yeah so we've been working on that, and adding platforms for bus and train stops etc
13:59 <novalis_dt> I think the new license is not very good.
13:59 <mele> yeah some people have refused it because they really don't agree with it too, and so their work will be lost
13:59 <kpw> why?
13:59 <mele> it's a shame
14:00 <novalis_dt> It specifies various specific EU laws rather than being something that would work everywhere
14:00 <novalis_dt> Some of it is barely English.
14:00 <mele> i think we're going to be losing most of the west coast's coastline -- should be fun
14:01 <mattwigway> I would assume that they would fall back to the TIGER import, but it looks like there are even some TIGER streets missing.
14:01 <mele> right things have definetly happened since the tiger import
14:02 <mele> and whenever someone (who didn't agree to the license) split a tiger street, that becomes something they created
14:02 <grant_h> mele: I heard that, but i know mapnik doesn't even use the coastlines in osm for their rendering (they use a separate shapefile
14:02 <mele> yeah i hope they figure out the right way to approach it, grant, i just don't really know that they'll have a handle on everything right away
14:02 <mele> i expect some chaos for a few weeks
14:02 <mele> hopefully i'm just being pessimistic
14:03 <mele> it's pretty bad timing though with all the good press osm has been getting lately
14:03 <mattwigway> RE: tiger, will the old TIGER streets be reinstated if those edits are reverted?
14:04 <novalis_dt> It is not necessarily possible to do that
14:04 <grant_h> I thik things will be fine, most major cities are in really good shape plus I think this whole process has and will encourage people to more actively map
14:04 <mele> mattwigway: i haven't been able to sort out exactly what they're going to do
14:04 <mele> yeah grant is generally more optimistic than i am :)
14:06 <novalis_dt> OK, so OSMpocalypse aside, anyone else have anything?
14:06 <grant_h> mattwigway: the details on what exactly is retained are a bit murky, but is a feature is created by somone that hasn't agreed to the new license it definitely gets deleted (and this the case with one half a of tiger way when it's split in two by a non-agreer)
14:07 <kpw> so team otp, let's keep our planet.pfb...
14:07 <demory> oh one quick thing
14:07 <demory> the deadline for railvolution submittals is tomorrow
14:07 <andrewbyrd> there was talk a few days ago about improved stop time interpolation, is that something you want to work on sooner or later?
14:08 <demory> anyone doing anything otp-related? we are planning to submit something focused on the analyst work
14:08 <novalis_dt> What systems does it affect?
14:08 <andrewbyrd> because I need to apply interpolation to repeated stop times, for at least one project
14:09 <novalis_dt> repeated stop times?
14:09 <andrewbyrd> kpw: looks like the current planet file will continue to be available "forever" under CC license
14:10 <kpw> oh! cool
14:10 <novalis_dt> Sure, but it will get more and more out of date.
14:10 <kpw> oh! right it forks
14:10 <kpw> which is bad
14:10 <andrewbyrd> novalis_dt: right, but we don't need to hurry out and buy drives to back it up on
14:11 <novalis_dt> Yep.
14:12 <andrewbyrd> well, wait... if one license allows incorporating changes from the other...
14:12 <novalis_dt> The problem is a bit complex
14:13 <novalis_dt> I don't think either really does grant all the rights the other does.
14:13 <kpw> demory: sorry, re railvolution
14:13 <kpw> let's submit in the innovation track
14:13 <kpw> i think the analyst work is perfect and we should be able to re-use some of our existing text
14:13 <kpw> do you want to work together on that in the am?
14:14 <andrewbyrd> novalis_dt: re repeated stop times, I have a route in SF where instead of leaving stoptimes blank for non-timepoint stops, they repeated the last timepoint over and over
14:14 <andrewbyrd> I thought it would be a reasonable default to treat repeated stop times as if they were blank in the GTFS and interpolate them
14:14 <demory> kpw: sure. I may put together a draft abstract this afternoon
14:15 <kpw> great!
14:15 <demory> we can probably borrow from some of the other analyst narrative we've written for other purposes
14:15 <novalis_dt> andrewbyrd, that is insane, but yes, that's a reasonable default
14:16 <andrewbyrd> kpw: do you guys need any visuals for the analyst paper?
14:16 <demory> andrewbyrd, i think they just want a text summary
14:17 <demory> although, we could point them to something online. maybe this is a good time to update that demo..
14:18 <demory> i may look into that today/tomorrow
14:18 <andrewbyrd> novalis_dt, this is the cause of the infinite speed warnings you will now see when building some graphs
14:18 <novalis_dt> andrewbyrd, I haven't really done anything in SF, so I haven't noticed.
14:19 <andrewbyrd> unfortunately I see infinite speed warnings in many graph builds
14:19 <novalis_dt> I see non-infinite-but-high speed warnings often
14:21 <andrewbyrd> they are annotated HOP_ZERO_TIME if the arrive/depart times match exactly, otherwise HOP_SPEED of infinity if the depart time is the same as the arrive time. 13000 of them in New York for example.
14:23 <novalis_dt> One problem with better interpolation is that it should probably rely on the street matching code
14:23 <andrewbyrd> right, that's why I was asking. I don't want to spend time on doing this fast, then just replace it later.
14:24 <novalis_dt> I'm really feeling like there's some kind of chicken/egg problem
14:25 <novalis_dt> Because I think having OTP spend a bunch of time fixing up bad GTFS is unfortunate
14:25 <novalis_dt> Since other tools might like better GTFS too
14:25 <novalis_dt> I think it makes more sense to have a generic GTFS fixing tool
14:25 <novalis_dt> Which I know KPW has been working on, but I'm not sure what the status is
14:25 <andrewbyrd> I know what you mean
14:25 <novalis_dt> That's one reason a I broke the OSM loading code out into a separate package
14:25 <novalis_dt> So that other tools could use it
14:26 <novalis_dt> My hope is eventually to remove the MapBuilder stuff from OTP and rebuild it outside with some tool that reads OSM and GTFS and lets you do things like snap stops to geometry and other such
14:27 <andrewbyrd> how far along is the basic GTFS fixer? I need to do this very soon, like tomorrow and this weekend, but it would be nice if it fit into a more generic tool.
14:28 <novalis_dt> You need better interpolation or just some interpolation?
14:29 <andrewbyrd> it would be nice if we used proper distance, but any interpolation would be better than giving trips with 0-second transit legs
14:31 <novalis_dt> Right, so I would say just hack in something that deletes duplicate times from StopTimes and then let the idiotic interpolator do the work
14:31 <novalis_dt> And at some point we'll figure out a roadmap for our transit editing suite