W dniu 30.03.2015 18:07, Mateusz Konieczny napisał(a):
Mapping street areas should not use [highway=*; area=yes] -
[area:highway=*] is much better.

That's exactly what was proposed regarding street areas:

http://wiki.openstreetmap.org/wiki/Proposed_features/Street_area#Tagging

but I wouldn't care so much for "better" scheme. More on this later.

highway=pedestrian, highway=footway for squares are special cases as
square may and typically is traversed using any route.

The catch is they can be just a more accurate micro-scale description of ways (with clear route) as well. And because micromapping (think zoom=19+) is making the progress, suddenly one day this will be the most prolific use of these schemes, not the squares!

As innocent as your statement sounds, some general tagging-related problems strike us badly again here:
1. Hard decision-making process.
2. Very limited navigation through the approved tagging schemes and changing them later.
3. Lack of general rules of recognizing tags for real objects.

***

Big red warning lights flash in my mind every time I hear about "special cases", because they make any systematic approach useless - or at least really hard. We wouldn't have the problem with natural=wood/landuse=forest if we care more about the tagging (eco-)system and sane generics than the Right Tagging Scheme for each case, because many of them are special in their own way and tagging schemes are very symbolic anyway.

So we have to decide each case separately:
1) area:highway=crossing or highway=crossing+area=yes?
2) area:highway=cycleway or highway=cycleway+area=yes?

- and we have very long and hairy discussions about such small details, which I frequently consider wasting too much of people's time.

In the meantime we forget about real rendering and routing issues with existing tools and the update path for them. How can we expect people to not tag for rendering if we don't care too much for such aspects while designing data structures they all (people and software) use?

***

Next we have to remember new schemes somehow - but since there's more and more of detailed cases with no common thread, it's getting almost impossible, so we have to rely on giant "phone book" (OSM Wiki).

Sadly, constant and detailed voting and documenting are now so complicated and so important that even Wiki is seen by many users of this list as too heavy (remember what "wiki" means?...)! I think Loomio is a nice tool to make deciding easier, especially if we have our own instance, but in my opinion there's just too much important voting/deciding in this model to be practical.

Now footway area, to our surprise, in general becomes more similar to streets than squares (because routing is more linear in most cases) and we have to deal with meaning shift or changing the tagging scheme to "better" one. Detailed documentation we relied on has to be heavy redesigned (more voting?) or starts to be misleading, which is sometimes worse than being too general.

***

Finally, we have to recognize and categorize objects from the real world - and we fail again, because we may have great wiki pages about railway=station and railway=halt, but they are so independent, that you have to think which features are defining for station/halt and which ones are just additional details.

We have no cascading/hierarchical categories to use in case of doubt (like, say, "railway=general_halt"), so we introduce cheating and mistagging - or we loose some informations, because mappers give up. That means taginfo statistics we use as a backing for proposals may be not so useful for finding what new schemes are really needed, because when people rely on ready schemes and have too few general rules, they are less eager to create something new and meaningful.

And once similar thing become popular among mappers (I don't know - hovercraft lines or whatever!), we will have to design new special cases, separate tags within "hovercraft=*" namespace (no "transport=general_halt" also, because "railway" key is our top-level namespace).

***

That's why I look for a next Mendeleev to bring us periodic table of tags, so we can see the order in the chaos we have today. And I think areas may be first to be redesigned in new, more clear and general terms.

--
Piaseczno Miasto Wąskotorowe

_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Reply via email to