[Talk-is] Footpath and cycleways...

[Christoph Hess]

Hi again.

Thank you for your fast reply.

> You are right, highway=path with access restrictions would be more
> correct according to this proposed feature on the wiki. However in
> practice the tools that actually use the data (renderers, osm->GPS
> exporters, routing applications) almost exclusively only look at the
> primary highway tag, i.e. whether something is a footway, cycleway,
> path, residential etc.

I don't agree with that. That Osmarender does a pretty good job with the 
path tag. See the example here around Háaleitis:

http://openstreetmap.org/?lat=64.13057&lon=-21.88755&zoom=16&layers=B000FTF

In this area all the ways I put in are highway=path and they have 
different access restrictions and rendered in different colors at least 
by the osmarenderer. And even if the mapnik renderer does not render it 
as a blue dotted line it is still correct and the data is correct. The 
path tag is quite common. And if it is a good tag and still proposed we 
should all vote for it... but I think it is already approved:

http://wiki.openstreetmap.org/wiki/Approved_features/Path

> And the cyclemap renders cycleways prominently while completely
> ignoring highway=path:
> 
> http://openstreetmap.org/?lat=64.1295&lon=-21.8819&zoom=14&layers=00B0FTF

You are right about this. But I don´t think that it is the best way to 
tag things as cycleways if they are not cycleways in reality. Then the 
renderer is just wrong and should be corrected.

http://openstreetmap.org/?lat=64.1295&lon=-21.8819&zoom=14&layers=00B0FTF

You see this is a small "mess" around Kringlan and the small bridge for 
foot and bicyclists. Now there is a way heading in direction Hvassaleiti 
and just ends. Because it is tagges as a cycleway. But who tells now in 
this bunch of paths what a cycleway or a pathway is? I think in reality 
this are just all paths and some could be used by bicycle and some not 
(like the small connecting to Hvassaleiti - but IMHO this is not a 
designated cycleway. To keep it according to reality we should use the 
path tag and correct the renderer and not squeezing it in so it is 
rendered nicely. Who decides if something is a cycleway when in reality 
there is nothing saying it is a cycleway?

> The same goes for the cyclemap I create for garmin devices, the
> stylesheet I use renders highway=cycleway very prominently:
> 
> http://wiki.openstreetmap.org/wiki/Garmin_map_of_Iceland

Then it should be no problem to use the path tag in your stylesheet, 
isn´t it? I never created one by myself but it think it should be not 
such a big deal. If I think on SQL it would be not so hard to add a new 
line that chooses data with highway=path AND bicycle=yes.

> My *only* interest in tagging conventions for the areas I edit has
> been to make them consistent, so that homogeneous objects share the
> same tags -- thus we can in the future mass-change them to any tagging
> scheme that fits us without loss of information, *and* to tag them in
> such a way that our maps are as useful as possible today for the
> widest possible audience.

I disagree here. We should try to tag it according to the reality and 
not according to the possibilities or mayby better to the limitations of 
the nowadays renderers. The renderers should be changed not the data. 
This will be a lot more work to change all the tags into the correct 
things then changing the renderer - there you can add few lines to 
correct the rendered map, but it is more work to change all the data 
(you need at least a smart program and a lot of computing time).

> For the latter reason I don't think we should change the
> highway=cycleways around the capital area to higway=path with
> appropriate access restriction. Even though it would mean the exact
> same thing as far as the raw map data is concerned it wouldn't as far
> as renderings of it go.

I was not talking about changing the data that is collected until now. 
It will be fine. But if someone puts in new data or is reworking some 
parts in the city he can maybe think about it.

> Any sane map rendering should include the cycle/footways such as those
> around Miklabraut more prominently than inner-suburb paths such as the
> ones in Háaleiti, but if we tagged everything as highway=path
> everything from a pavement to wide paved cycleways such as the ones in
> Grafarvogur would look exactly the same on the map.

Depends on the renderer, but the data is correct anyway. And it is still 
useless to separate a street from a sidewalk: it is a structure that 
belongs together. It belongs to the street and everybody knows that 
there is (in 95%) a sidwalk in a city. If the sidewalk is very far away 
(maybe 5 m) and it is not clearly stated that it belongs to the street I 
agree that it should be included but not every crossing of the sidewalk 
at a big crossing. I tried this before around this area and it is a big 
mess and it is just not according to the reality because there are 
features missing like all the different lanes on the road. I think here 
we are limited by the data not only by the renderer. And it would be 
useless at all because the good GPS devices are still way too 
tolerant... I think the more natural way would be to include information 
about sidewalks with the street tag as I stated. I think there is a big 
need in here and a lot of people want to include this information. But 
using more and more superfluous data does not help and may confuse the 
renderer more and more. When the way and sidewalk is too close it will 
be not rendered anyway because it will overlapping with the street 
rendering.

Hopefully there will be a lane tag and a sidewalk tag sometime. Even to 
put in big roads like Miklabraut (or Autobahn n Germany) with two 
separated roads is not perfect and sometimes rendered 
incorrectly/overlapping. It would maybe be better to put in one way and 
render it as two lanes according to a lane tag (a lot of people 
disagreed on this feature because it would be to complicated but maybe 
it would be better. Hey, who said reality is simple or not complicated ;-)

http://wiki.openstreetmap.org/wiki/Relations/Proposed/Lane

> I think I made my opinion at least pretty clear in my November post to
> the mailing list on the matter. I think if there's a feature on the
> ground we should map it and add it to the database. Adding a tag to
> the way saying there's a pavement on the left or right doesn't carry
> the same information as mapping out the pavement itself, it doesn't
> allow you to specify where a crossing occurs, how far the pavement is
> from it and where other footways intersect with that pavement.

Maybe we don't need to make it too complicated. If there is a footway 
crossing a street - most people will realize that it is mostly possible 
to walk along the street. And if there is a sidwalk tag in the future 
the interpreter of the data can exactly render it like you need it. And 
IMHO this is more natural and according to reality then having more 
streets and lanes and streets: It is not correctly at all to tag 
Miklabraut with two separated streets. Would it not maybe more correct 
to use 5 or sometimes six streets because we have 2 lanes in each 
direcion and a bus lane or a emergency lane?

> I don't see how mapping a different kind of way closely to another way
> is going to be confusing, it's not ambiguous in the database if tagged
> correctly, and if someone doesn't want to see or render it it's very
> easy to filter it out before using the data.

Rendering and parsing to much superfluous data is time consuming but 
tagging smart and rendering smart is time saving, less mistakes can 
happen and less inaccurate.

We should not only collect data according to the nowadays renderer but 
to keep an eye on the future things. The renderer will always be changed 
and it is very easy to change rendering. But correcting collected data 
is a major and time consuming process. I do not use a lane tag or 
sidewalk tag until now but anyway if there is no feature like a sidewalk 
i does not make sense at all to collect a lot of data about normal 
sidewalks beside the streets.

But, let´s discuss this more. Maybe we find maybe not the best but a 
better solution together.

Maybe see you tomorrow, I'm not totally sure if I can come tomorrow.

Christoph