forked from opentripplanner/OpenTripPlanner
-
Notifications
You must be signed in to change notification settings - Fork 3
2011.12.01 Weekly Check in
demory edited this page Dec 29, 2011
·
1 revision
- 13:30 <demory> Hello everyone.
- 13:30 <andrewbyrd> Hi
- 13:30 <demory> Shall we get started?
- 13:31 <novalis_dt> sure!
- 13:31 <demory> OK!
- 13:31 <demory> My update: checked in initial work on multi-destination routing last week, supporting selection of intermediate places from the map UI. I plan to return to this in next week or so to add geocoding support.
- 13:31 <demory> Otherwise, have been focused on non-Trimet work -- new community website, and discussion about analyst extension, new web-based debugger/visualizer, and hosted deployment manager. we can discuss those further later
- 13:32 <demory> That's all I have for now
- 13:32 <novalis_dt> Andrew and I have merged his graphvertex-refactor; I've been working on the web-based debugger/visualizer backend as well as json improvements.
- 13:33 <demory> FrankP grant_h mele: anything on the Trimet end?
- 13:34 <andrewbyrd> I just checked in a basic implementation of GraphBuilderAnnotations and am going through the GraphBuilders replacing log entries with annotations.
- 13:34 <grant_h> we've been working on a new weighting scheme that favors bike infrastructure even more
- 13:34 <grant_h> we'd like this to eventually replace the weights that are in place now
- 13:35 <grant_h> and if possible the old weights would be coupled with elevation and become a drop down option called "most efficient" or something like that
- 13:36 <novalis_dt> grant_h, can you explain that idea a bit more?
- 13:37 <grant_h> we've gotten feedback from users that they want to be directed to multi-use paths like springwater and the esplanade more readily than they are now when they chooose "safest"
- 13:37 <grant_h> sure
- 13:37 <mele> basically we want to offer an even safer option, but we still like the existing weights and think they still work for a lot of people so we want to keep those in some way as well
- 13:38 <novalis_dt> Wouldn't somewhere on the triangle between safest and fastest be more like the current weights?
- 13:38 -!- Peejay_ [a867e9a0@gateway/web/freenode/ip.168.103.233.160] has joined #opentripplanner
- 13:39 <mele> yeah, maybe somewhere I suppose but they don't scale right to have the right effect
- 13:39 <mele> i think before we were thinking of the current + elev as a "smartest" drop-down
- 13:39 <grant_h> probably, but it might be nice to have a drop down option so users don't have to repeatedly find that spot on the triangle if they prefer those weights
- 13:39 <Peejay_> except not using the word "smartest" because that might offend ppl
- 13:40 <mele> oy i think it's fine but whatever :)
- 13:40 <Peejay_> safety nerds won't like it!!!
- 13:41 <novalis_dt> OK, we'll probably want a "custom" routing option anyway
- 13:41 -!- mele_ [d819d1be@gateway/web/freenode/ip.216.25.209.190] has joined #opentripplanner
- 13:41 <novalis_dt> So we'll just declare the old weights to be "custom" weights. We can reuse custom for what guillarme barreau was doing with NO2 emissions
- 13:41 <novalis_dt> andrewbyrd, does that sound right?
- 13:42 -!- grant_h_ [d819d2bc@gateway/web/freenode/ip.216.25.210.188] has joined #opentripplanner
- 13:43 -!- FrankP [d819d177@gateway/web/freenode/ip.216.25.209.119] has quit [Ping timeout: 265 seconds]
- 13:43 <andrewbyrd> Yes, that seems good.
- 13:44 -!- grant_h [d819d2bc@gateway/web/freenode/ip.216.25.210.188] has quit [Ping timeout: 265 seconds]
- 13:44 <novalis_dt> OK, I can add that quickly, if Frank gives me the go-ahead.
- 13:44 <novalis_dt> Do you guys have anything else?
- 13:44 -!- FrankP [d819d2f0@gateway/web/freenode/ip.216.25.210.240] has joined #opentripplanner
- 13:44 -!- mele [d819d1be@gateway/web/freenode/ip.216.25.209.190] has quit [Ping timeout: 265 seconds]
- 13:44 <grant_h_> Another feature that I think would be nice is providing multiple options when users plan bike only trip, so we could run both sets of weights and list two options, would that be diffcult to implement?
- 13:45 -!- Peejay_ [a867e9a0@gateway/web/freenode/ip.168.103.233.160] has quit [Ping timeout: 265 seconds]
- 13:46 <andrewbyrd> Not necessarily difficult, just different.
- 13:46 <andrewbyrd> Multi-objective search offers the possibility to find the N "best" routes, considering safety, hills, and speed separately, at the same time.
- 13:46 <demory> would these be preset weights (safest, etc.) or custom ones from the triangle?
- 13:47 <andrewbyrd> Trying that out is just a matter of replacing a comparison function.
- 13:47 <demory> (or perhaps a combination of both?)
- 13:47 <novalis_dt> demory, the alternate weights is more about the individual road weightings, rather than the mix.
- 13:47 <demory> oh ok
- 13:48 <andrewbyrd> As currently implemented it would be a combination of both: trips would be ranked overall by the triangle, but different extremes of safety/speed etc would all be taken into account.
- 13:48 -!- Peejay_ [a867e9a0@gateway/web/freenode/ip.168.103.233.160] has joined #opentripplanner
- 13:48 <novalis_dt> Of course, this is all dependent on us getting multiobjective routing working, which it's presently not.
- 13:49 <andrewbyrd> Trying several different road weighting schemes or triangle positions to get multiple itineraries would mean going back to repeatedly re-routing.
- 13:50 <novalis_dt> But you could do it without redoing the bidirectional calculation
- 13:50 <novalis_dt> because the lower bound on the backwards search is independent of optimization criteria
- 13:50 <novalis_dt> (for now)
- 13:51 <andrewbyrd> It is for now, and that might need to be changed if you want multiple bike routes at one go (from a single search).
- 13:51 <andrewbyrd> Good things to try out at any rate, and see if they provide good variety in results.
- 13:52 <demory> anything else relating to TriMet?
- 13:52 <FrankP> My update: Printing is done (but I'd like to improve it, so ticket stays open); Also put a switch in front of the multi-destination stuff, so it can be turned off (plan to release a new RTP on the 15th ... multi destination stuff turned off).
- 13:54 <demory> ok so the 12/15 version will not have multi-destination enabled by default?
- 13:54 <novalis_dt> FrankP, do you want me to go ahead on providing the custom weighting scheme proposed by the interns?
- 13:54 <FrankP> OTP has it enabled...but for our RTP instance, I've disabled it.
- 13:55 <demory> ok. prob a good idea, it could use a little more work
- 13:56 <FrankP> novalis_dt, I don't think you should work on TriMet's behalf w/out getting approval from Bibianna (as mentioned earlier, she's looking to preserve a few hours). That said, for the betterment of the project, don't let me stop ya ...
- 13:56 <novalis_dt> FrankP, well, I have no shortage of work for the betterment of the project.
- 13:57 <novalis_dt> So, interns, if you guys want the feature, lobby Bibiana :)
- 13:57 <FrankP> true...(especially that guy leen has been quite insistent on your time)
- 13:58 <grant_h_> Sounds good, also we just got these new weights up and running on a test instance of OTP so we should probably test and tweak them a bit more
- 13:58 <novalis_dt> FrankP, yeah, but his problems are really quick to ansewr. No, it's really Kevin who has been insistent :)
- 14:00 <demory> speaking of betterment of the project... that's a good segue to updates on non-TriMet work
- 14:00 <demory> unless you had anything else?
- 14:01 <demory> ok
- 14:01 <demory> analyst extension: David T, Andrew, Kevin and I spent some time this week (we're all in NYC) mapping out an initial phase of work on a more robust version of the travel-time tools. After this week I will write up a summary and post to wiki/list
- 14:01 <demory> visualizer/debugger front-end: starting work on new web UI (probably leaflet-based) to allow for visualization of graph elements and graph-building errors, using new API functions currently being written (David T checked one in yesterday). envisioned as successor to current java visualizer
- 14:02 <demory> hosted deployment: David T has completed initial phase of work on GTFS submittal UI / validator / transloader; Kevin and I are working on graph-builder automation this week as well. hope to have this ready in time for TranspoCamp DC in Jan
- 14:03 <demory> any comments/questions?
- 14:04 <demory> ok well that's all i had
- 14:04 <demory> anything else?
- 14:04 <novalis_dt> not here.
- 14:05 <demory> so next week, i'll likely be speaking at the GA Transit Association conference during this time
- 14:06 <demory> you all want to go ahead and do the call w/o me?
- 14:06 <novalis_dt> Sure!
- 14:06 <demory> (i may be able to join depending on the specific schedule for the afternoon)
- 14:06 <demory> ok great
- 14:07 <demory> i'll send out a reminder thurs in either case, should know more about schedule then
- 14:07 <demory> have a good week!
- 14:08 <andrewbyrd> Thanks. Bye everyone.