2013.04.04 Weekly Check in - atlregional/OpenTripPlanner GitHub Wiki
13:32 <demory> hey FrankP, grant_h
13:33 <mattwigway> hi
13:33 <FrankP> hey
13:33 <demory> and mattwigway, sorry!
13:33 <mattwigway> np
13:34 <demory> ok, let's do a quick check-in
13:34 <demory> abyrd had something else tonight (it's evening there) so i don't think he's actually on
13:34 <demory> but i do have a quick update from him
13:35 <demory> he and i have been looking into #945 and the other TriMet-reported bugs
13:36 <demory> we have been able to reproduce the #945 examples in the last error report you sent us grant_h
13:36 <grant_h> great!
13:37 <demory> abyrd thinks it has something to do with walk limit handling
13:37 <demory> note that in those cases if you raise the walk limit, the "right" result is returned
13:38 <grant_h> I wouldn't be surprised, I think most of the issues we've been having relate back walk limits
13:38 <mattwigway> I'm just starting some new research using Analyst for accessibility analysis.
13:38 <demory> so he's going to keep looking into that. but I think we at least have an idea of what's causing the problem
13:39 <demory> mattwigway, great
13:39 <demory> there's a lot of interest in Analyst right now
13:39 <demory> we had a good talk w/ Jarrett Walker (of Human Transit fame) this week about some partnership opportunities there
13:40 <demory> i'll keep you posted if anything comes of that
13:40 <mattwigway> WRT that, I'd mention #990 as something to be aware of
13:40 <FrankP> The graph and code is being built every night from head ... there's an interesting trip from Sunset TC to Zoo, which with 1/2 mile walk, the trip is strange ... but increase the walk to 3/4, and things do look better -- but would still need that 1/2 mile walk to do something much better than what's happening now -- customers will get very confused over that type of result.
13:40 <mattwigway> It appears that something broke analyst transfer compression at some point.
13:41 <mattwigway> I haven't had a chance to look into it (and it may be a while before I do)
13:41 <demory> ok, thanks, we'll take a look when we get the chance
13:42 <demory> resolving the TriMet-reported routing issues are first priority right now
13:42 <mattwigway> if you get to it before I do, demory, feel free to contact me with questions. the code is somewhat complicated.
13:42 <FrankP> BTW, that Sunset to Zoo trip works fine with a 1/10 mile max walk -- just 1/4 & 1/2 show a problem.
13:42 <demory> otherwise, i've also bringing myself up to speed on the TransitIndex code, will be expanding that API some to support UI enhancements
13:43 <demory> hmm, interesting
13:43 <demory> so, clearly the walk limit has something to do with it
13:44 <mattwigway> It could be a doubling thing: 1/10 * 2^n can never be 1/2 (or 1/4)
13:44 <mattwigway> i.e. it goes 1/5, 2/5, 4/5, 8/5 and so on.
13:44 <mattwigway> whereas 1/4 -> 1/2
13:45 <mattwigway> I wonder if 1/8 shows the problem
13:45 <mattwigway> because it could be something happening with almost exactly 1/2 mile walk