remove line barrier catch-all rendering

[nebulon42]

Removes line-barrier catch-all rendering and fixes a bug where the line-barrier layer was displayed from z16, but a rendering rule for z14 was present.

Line barrier values > 1000 uses (with taginfo usage and comments):

fence 1 748 621 ok
wall 1 306 878 ok
hedge 646 979 ok
retaining_wall 93 566 ok
yes 47 152 no wiki page, too general
kerb 44 915 ok
ditch 20 143 ok
city_wall 19 589 ok
guard_rail 17 077 ok
hedge_bank 14 508 only german wiki page available
wire_fence 10 379 no wiki page, should be fence
line 8 166 no wiki page, not a barrier (access=no?), tagging for the renderer
chain 5 405 ok
embankment 4 599 no wiki page, but included since rendering rule already present
field_boundary 3 847 no wiki page
wood_fence 2 171 no wiki page, should be fence
avalanche_protection 1 865 no wiki page
handrail 969 ok

[matkoniecz]

fixes a bug where the line-barrier layer was displayed from z16, but a rendering rule for z14 was present

From reading code - this PR adds rendering of barrier lines on z14 and z15. I am guessing that it will give poor results at z15 (maybe except remote areas) and nearly always will make map worse on z14.

Also, what are before/after results? Especially on z14 and z15.

[nebulon42]

this PR adds rendering of barrier lines on z14 and z15

https://github.com/gravitystorm/openstreetmap-carto/blob/master/landcover.mss#L628 should prevent that, or am I overlooking something?

One more note: Removal of barrier=line is likely to upset mappers who like to map sports field markings, but I think this is tagging for the renderer and should not be encouraged.

[trigpoint]

Field_boundary is used when the boundary is not easy to determine, usually fence or hedge.
Sometimes it is used when a boundary frequently changes.
The primary purpose is to map the boundary, as barriers are the most important feature when mapping the countryside, the detail doesn't really matter.

[matkoniecz]

https://github.com/gravitystorm/openstreetmap-carto/blob/master/landcover.mss#L628 should prevent that, or am I overlooking something?

So only barrier = embankment will be now rendered on z14 and z15. But it is worth checking is it a good idea.

One more note: Removal of barrier=line is likely to upset mappers who like to map sports field markings, but I think this is tagging for the renderer and should not be encouraged.

It is undocumented and weird - the only reason to use barrier as main key is tagging for renderer.

Field_boundary is used when (...)

It would be far better to document it on wiki and link it here.

Field_boundary is used when the boundary is not easy to determine, usually fence or hedge.

So it is basically equivalent of barrier=yes.

[matkoniecz]

I would consider adding also http://taginfo.openstreetmap.org/tags/barrier=handrail
(used 969 times, documented on wiki).

Maybe also other values with 100 and more usages (like it was done for shops).

[matthijsmelissen]

Looks good to me at first sight.

[imagico]

Right now barrier, in particular barrier=ditch in combination with waterway tags looks strange and non-intuitive, like a waterway with a different color:

http://www.openstreetmap.org/#map=19/48.00544/7.78167

Maybe you can change this PR to not render barrier=ditch when in combination with a waterway tag that is rendered (i.e. stream/drain/ditch).

[nebulon42]

So only barrier = embankment will be now rendered on z14 and z15. But it is worth checking is it a good idea.

I cannot test this as the only occurrence in my data is covered by a road at lower zooms. But since the code was in I assumed that it was ok once. Maybe somebody else can confirm.

I would consider adding also http://taginfo.openstreetmap.org/tags/barrier=handrail

added, 100 times is a bit low, but if there is a consensus about that I'm ok with it

Maybe you can change this PR to not render barrier=ditch when in combination with a waterway tag that is rendered

Interesting, never seen this before. Added an exception for waterways.

[polarbearing]

Values less than 1000 should be on consensual selection, without numerical limit. Thus useful new values can be added, while wrong spelling and tagging for the renderer can be suppressed. Just created a deprecation page for barrier=curb (595) as an AE spelling of kerb (44915).

[dieterdreist]

2015-11-29 19:57 GMT+01:00 Christoph Hormann [email protected]:

Maybe you can change this PR to not render barrier=ditch when in
combination with a waterway tag that is rendered (i.e. stream/drain/ditch).

+1. Any waterway will be in a depression (or in a man made tube / "open
tube"), so this combination seems weird anyway. If there is a significant
cutting or lateral embankment/slope, it should be mapped like that.

[dieterdreist]

yes 47 152 no wiki page, too general

yes, it's generic (and though not too helpful without further information)
but is OK IMHO (as a preliminary mapping).

hedge_bank 14 508 only german wiki page available

how does this relate to barrier=hedge?
IMHO we should generally develop a way to map barriers that consist of
several layers (e.g. retaining wall, above a fence, etc.), maybe with a
linear relation (to reuse the geometry and avoid drawing overlapping ways)
and layer tags.

wire_fence 10 379 no wiki page, should be fence

+1

embankment 4 599 no wiki page, but included since rendering rule already
present

what's the difference between this and a man_made=embankment? If it's very,
very steep, it will be a retaining wall or a cliff, otherwise it is an
embankment like any embankment, not?

wood_fence 2 171 no wiki page, should be fence

+1

[HolgerJeromin]

I dont know if everybody is able to read the german wikipage to barrier=hedge_bank.
It is kind of barrier=hedge+embankment=yes.

[nebulon42]

There was an error in the SQL related to waterways, which I have fixed.

[matthijsmelissen]

Thanks!

[nebulon42]

@math1985 @pnorman @matkoniecz @gravitystorm I plan to inform the community (especially the German one) beforehand of this change to minimise a probable backlash. When a good tagging scheme for e.g. barrier=line is established re-adding sports field markings to this style should be no problem.

Objections, thoughts?

[imagico]

My opinion is that explicit mapping of sports field markings is not a good idea and should not be encouraged by renderers. The problem is of course that the alternative, i.e. rule based rendering of markings based on a generic sport type tagging of fields (#1126) is complex to implement within the current framework here.

[matthijsmelissen]

I plan to inform the community (especially the German one) beforehand of this change to minimise a probable backlash.

Thanks, good idea!

When a good tagging scheme for e.g. barrier=line is established re-adding sports field markings to this style should be no problem

I think the discussion whether or not to map field lines is out of the scope of the repository. We could discuss whether we want to render them, independently of the tagging scheme. I suspect the answer to that question would be 'no'.

[matkoniecz]

I plan to inform the community (especially the German one) beforehand of this change to minimise a probable backlash.

Great idea, thanks for doing it!