[Talk-us] Updating tagging of public transport

[Greg Morgan]

On Tue, Feb 3, 2015 at 1:31 PM, Paul Johnson <baloo at ursamundi.org> wrote:
> Is the new style backwards compatible with the old style?  Especially in
> regards to potentially having OSM as being the dataset for the official GTFS
> feed in an area where the official data presently sucks, it'd be a travesty
> if the new style makes going to and from GTFS more difficult for marginal
> (and primarily German micromapping) gains.

I am in the process of implementing the Valley Metro, VM, bus system
in the Phoenix Metro area.  You can see the progress here
http://www.openstreetmap.org/#map=15/33.4527/-112.0688&layers=T

This wiki page provides a todo list and status information.  I am
resting on the shoulders of giants by using the original relations
created by other mappers as the route masters.  There was some work
involved in generating the wiki page but it is nice having a spec and
status page to work with
http://wiki.openstreetmap.org/wiki/Arizona/Maricopa_County/Public_Transport

Some of the design thoughts are in the talk page. I plan to add my
experiences along with procedures on the talk page. I've been taking
screen shots and notes as I go along.  I did not know you could do
some of the things with JOSM that help implement the routes. Hopefully
having a blow by blow how to will help others make the switch to the
new tagging scheme or implement their bus system in OSM.  Perhaps some
of the information can help with the chicken and egg issues that
Saikrishna mentions. I didn't find real good implementation plans in
the wiki other than similar status pages.  Missing howtos may be a
larger problem than the complexity of the tagging scheme.  As a
volunteer it is easy to through out a bus stop node; save; and run an
errand. Pre-planning and leaving a bread crumb path for other mappers
is not normally how I map.  I imagine some of the European mappers
worked the same way
http://wiki.openstreetmap.org/wiki/Talk:Arizona/Maricopa_County/Public_Transport

One of the problems is the 24 to 48 hour rendering cycle (at least is
feels that way) of the transportation map.  Even though you think that
you have it right, there are still mistakes that are made.  That's OK.
The maps is getting better.  I also want to thank Andy Allan for
producing the map.  It would be great if he could respect the colour
tag for each route but then again I think that I understand the
limitations of the current rendering systems.  Rendering the routes of
light rail systems may be a useful change for the transportation map.

As far as the legacy question, I had to add this park and ride/transit
center location for the 44 route
http://www.openstreetmap.org/#map=18/33.60488/-111.97937&layers=T.  I
used the legacy tag to mark the station for now.  I used
landuse=industrial for now but I know that is not the correct long
term tag.  It helps me find locations during mapping. I believe a
mixture of the two tagging systems will co-exist. The tags can be
adjusted later.  I've already made several passes through the
relationships to fix the tags.

The routes weren't too bad to figure out a plan but I started mapping
when my eyes glazed over for the stops.   Part of the plan is to
figure out stops when the routes are done. Using a ref/name of the bus
line will not be as useful as using the VM Next Right number for the
stop. Hans has several of these mapped per this area and stop
http://www.openstreetmap.org/node/2253674372.

Right now a big win is having data to talk about verses a blank page.
The transportation map shows something.  I see the need to reorganize
the wiki page more to a local and regional route organization similar
to the local regional value of network tag.  Currently the wiki
organization mirrors more of VM list of routes.  However, all the VM's
rapid, express, and limited stop routes feel more like regional
routes.  I can imagine how big of a project that SteveA's UBRS bike
route project was/is.  Oh my goodness the metro area is big!  I have a
feeling that bus routes are more challenging that train or national
bike routes because you have to plow through all the connected ways of
an urban setting.

The nice thing about the relationship is that I had the temporary
route changes for the downtown Super Bowl Experience mapped.  I
removed those last night so they will soon disappear.  Slurping up
existing ways to make the route or route change sure is nice.  LOL I
would have had the whole system done for the Super Bowl but some other
interesting areas caught my eye.

All the routes in waiting have been assigned to the ValleyMetro
building here http://www.openstreetmap.org/way/30704302. Templating
the routes with tags has been a big plus. Using one way for holding
the routes has been very useful. Templating one route that many other
bus lines share was a great time saver.  Vim was handy at times during
the templating process i.e. make a JOSM relationship; save the data to
a .osm file; yyp in vim to create and edit other routes.

Finally, I am thankful that MapBox made their updated satellite images
available.  The imagery helped me complete some of the newer park and
ride/transit centers. It would be nice if there was a version of their
imagery that would let me over zoom the last level to 19, 20, and 21.
The pixelation version would be useful in creating some of these newer
areas.  Currently, it looks like Bing pops in for the detailed
imagery.  I am looking for an excuse to go GPS one transit areas that
is not in the MapBox imagery.  ;-)

HTH,
Greg

>
> On Sun, Dec 7, 2014 at 2:21 PM, Saikrishna Arcot <saiarcot895 at gmail.com>
> wrote:
>>
>> Part of the problem between the tagging schemes and the rendering is that
>> it's a chicken-and-egg problem; a new tagging scheme is created, but
>> rendering support isn't there yet (partly because it's a somewhat complex
>> structure), so people might not use that scheme. However, if there were many
>> instances of using the newer scheme, then it would be justified for the
>> renderers to add support for that scheme.
>>
>> (On the rendering topic, though, I can confirm that OSM's transport map
>> does support the newer scheme, as does Öpnvkarte, OpenStreetBrowser, OsmAnd,
>> so it's not lacking.)
>>
>> A slightly bigger issue I see is that there are two formats for tagging
>> transportation routes, which will not only require data consumers to code
>> for both formats, but will also make it harder to link a bus route tagged
>> using the newer format be "connected" to another bus route using the older
>> format. I feel that this should be resolved quickly.
>>
>> On Saturday, November 29, 2014 02:02:01 stevea wrote:
>> > >it is not clear if the new way is actually better, at least the
>> > >current data stats show that mappers still prefer the "old" method,
>> > >at least for bus stops, as it is simpler (you need just one tag
>> > >highway=bus_stop instead of two: public_transport=platform and
>> > >bus=yes, for the same information content), and the new style cannot
>> > >be rendered on the main map, because of the lack of the bus-key (the
>> > >rendering db only "knows" that there is some kind of stop, but it
>> > >cannot determine if it is a tram stop, a bus stop or whatelse).
>> > >
>> > >I wouldn't "re-tag", ie. won't remove tags, but you can add the
>> > >public_transport=* tags if you want to support also this scheme.
>> >
>> > Is what I hear Martin saying here is that tagging with an old style
>> > because it renders AND tagging with a newer syntax that doesn't is
>> > OK?  (As in, "doing two things at once, even if they achieve
>> > different, but good and worthy goals, is right"?)  If so, part of
>> > what it says is that syntax is rather distantly connected to
>> > rendering.  Read that again, as I think it is important.  It is about
>> > what might be called OSM's "transmission."
>> >
>> > Not everybody understands the full process of how changes in syntax
>> > (e.g. voted upon tagging) turn into "what we see mapped."  There are
>> > human consensus processes there, there are coding processes there
>> > (including bug fixes, actual writing of render code..) there is quite
>> > much more than just that there.  It is a complicated moving set of
>> > parts.  It is "let's map bus routes, OK, let's describe better syntax
>> > for bus routes, OK (but we don't render that today").  Now what?
>> > That's a real "hit the brakes and think about how to do it better, so
>> > discuss" moment.
>> >
>> > As we recognize distance between what people want to see represented
>> > in the map (how they tag) with the syntax of doing so (actual tags
>> > that get into OSM's data) can we better discuss this?  We can and
>> > should, I say.  Deep, I know.  My point is that a person wanting to
>> > understand how to influence this is very much helped by understanding
>> > it (as much of it as possible, as much of it as we can describe as
>> > what we intend...) in the first place.  How might one see such moving
>> > parts of OSM and how they a) work today? and b) work better in the
>> > future as we intend them?  It goes deeper than public transport
>> > tagging, but that is a good example through this transmission.
>> >
>> > Look, I know:  some of us work on our transmission, and they must.  A
>> > lot more of us -- and there are many -- are only quite vaguely aware
>> > of how it works, or how we might best induce positive change into its
>> > workings.  We can do better.  Good discussion so far, but it seems we
>> > are only scratching this surface.
>> >
>> > SteveA
>> > California
>> --
>> Saikrishna Arcot

>