[Talk-transit] All about bus stops

Gerrit Lammert osm at 00l.de
Thu Jan 22 19:55:59 GMT 2009

Hi Gerald.

Gerald A wrote:
> --------------------------------
>   <   <   <   <   <   <   <
> --------------------------------
> ###<###<###<###<###<###<###<####
> ================================
> ###>###>###>###>###>###>###>####
> --------------<00000000000000>---------
>   >   >   >   >   >   >   >
> --------------------------------
> Your "tram" stop floated in the middle of the tram line, but people were
> waiting on the curb. The "<OO>" bit I added represents the island I was
> talking about.
> Our streetcars have right of way, and without one of these islands,
> passengers getting on and off would stop the traffic in the ">>>" lane by
> law. The link I sent you shows an image of one of these, and they are almost
> always near a cross-street with traffic signals.

OK, this is actally the exact situation I was referring to the first 
time. I guess our google maps examples were not good enough. :)
In mine the rails are barely visible and there are now landmarks on the 
"island". You can see a couple of people on the right side though, 
indicating there might be something in the middle of the road to stand 
upon. :)
Back to the scenario. stop tagging scheme aside:
When I encounter a piece of way that is physically devided (like by the 
island in this example), I would split the road/tram-combination and 
have them modelled as different lanes for this short section.
In the above example, I would have 3 lanes.
One combining the uppermost lanes (highway=xxx;railway=tram;oneway=yes)
One for the tram in the other direction (railway=tram;oneway=yes)
And one for the missing road lane (highway=xxx;oneway=yes).
I would do this, because othe than before and after the "island" i could 
not physically cross the opposite lane from my line.
IF this situation is mapped in this way, the bus stop and 
platform/shelter/... positions can be placed at the proper positions.

for the earlier mentioned real life example.

>> In general: There are a lot of situation with trams and other vehicles than
>> can not be mapped 100% accurate by now.
> Agreed.
>> Some propose to always have a discrete way for every "lane" like:
>>     |      |           |      |           |         |           |
>>     |      |           |      |           |         |           |
>> foot | bike | curb/kerb | car1 | car2&tram | barrier | car2&tram | ...
>>     |      |           |      |           |         |           |
>>     |      |           |      |           |         |           |
>> But I rather loose some detail (like: on a two lane-road with combined
>> tram, is the tram on the first or second lane?) than mess with multiple ways
>> and all the added information on where one is able to change from one to the
>> other...
> Well, this was sort of what I was getting at -- without creating a mapping
> nightmare in terms of mapping individual lanes as ways around this one
> feature, how can we take this into account?

See above. In general I'm very undecided on how to go at these things. 
In the end dividing into different lanes might be the most general 
solution, but I dread the editing nightmare and the data amount 
explosion. That's why I settle for this hybrid approach...

> I hope all is clear now ... and thanks for talking it out with me. Makes it
> a bit clearer for me as well.

Thats the idea. Charing best practices. :)


PS: Sorry for this incomplete discussion. I always forget to add the 
mailing list when hitting reply. Could someone modify the list's 
settings so that the mailing lists address is the default reply-to?

More information about the Talk-transit mailing list