[PATCH] Add support for crossing of residential areas with something.

[marxin]

The patch makes detection of area more generic and adds support for residential areas.

[Don-vip]

Thanks for the patch!
I see some issues to fix before merging:

• the removal of mapcss rule means there are several use cases no longer detected (for example, two crossing industrial landuses). It should rather skip the residential value.
• the translations using string concatenation (tr("Crossing " + types[0] + "s")) won't work. The text to translate is statically extracted from source code. Full strings must be used.

[marxin]

Replying to Don-vip:

Thanks for the patch!
I see some issues to fix before merging:

• the removal of mapcss rule means there are several use cases no longer detected (for example, two crossing industrial landuses). It should rather skip the residential value.

Can you please help me how to exclude landuse=residential in the mapcss. I'm not much familiar with that.

• the translations using string concatenation (tr("Crossing " + types[0] + "s")) won't work. The text to translate is statically extracted from source code. Full strings must be used.

Oh, you are right. I'm sending updated patch, but it doesn't scale much as one needs to explicitly handle all N*(N-1) possible values.

[Don-vip]

Replying to marxin:

Can you please help me how to exclude landuse=residential in the mapcss. I'm not much familiar with that.

Probably something like that:

#mapcss
area:closed:areaStyle[landuse!=residential][tag("landuse") = parent_tag("landuse")] ⧉ area:closed:areaStyle[landuse!=residential]

MapCSS expert in copy :)

[Klumbumbus]

OK, now that you added me in keywords and cc you got my attention ;)

Replying to Don-vip:

Probably something like that:

area:closed:areaStyle[landuse!=residential][tag("landuse") = parent_tag("landuse")] ⧉ area:closed:areaStyle[landuse!=residential]

Yes, simply add [landuse!=residential] to exclude this tag, but I think you forgot/overwrote the last [landuse] at the end of the selector:

area:closed:areaStyle[landuse!=residential][tag("landuse") = parent_tag("landuse")] ⧉ area:closed:areaStyle[landuse!=residential][landuse]

[Don-vip]

Replying to Klumbumbus:

OK, now that you added me in keywords and cc you got my attention ;)

Haha sorry I often do the mistake.

I think you forgot/overwrote the last [landuse] at the end of the selector:

This was intentional. [landuse!=residential] implies we have a landuse value, different from residential, right?

[Klumbumbus]

Replying to Don-vip:

[landuse!=residential] implies we have a landuse value, different from residential, right?

No, e.g. way[landuse!=residential]{} matches a way tagged only with natural=wood and an untagged way too

[marxin]

So can you please provide a final version of the mapcss change.
I'm bit confused which selector is the right one?

[Klumbumbus]

If I understand the changed Java test right, then the mapcss selector should change to:

area:closed:areaStyle[landuse!=residential][tag("landuse") = parent_tag("landuse")] ⧉ area:closed:areaStyle[landuse!=residential][landuse]

[marxin]

V3 of the patch

[marxin]

Replying to Klumbumbus:

If I understand the changed Java test right, then the mapcss selector should change to:

area:closed:areaStyle[landuse!=residential][tag("landuse") = parent_tag("landuse")] ⧉ area:closed:areaStyle[landuse!=residential][landuse]

Thanks for that. I verified it works for e.g. industrial landuses. Due to that I found one issue in my patch:

p.get("landuse") == "residential"

is different from

p.hasKey("landuse", "residential")

I'm attaching new version of the patch.

[Don-vip]

Replying to marxin:

p.get("landuse") == "residential"

is different from

p.hasKey("landuse", "residential")

Yes the good call is:

p.hasTag("landuse", "residential")

p.hasKey("landuse", "residential") means p has landuse=* or residential=*

[Don-vip]

In 13671/josm:

fix #16188 - Detect crossing of residential areas (patch by marxin, modified)

[Don-vip]

Thanks for the patch! I have modified it a little bit to replace the string by an enum. I didn't test it much so I hope I didn't break anything.

[Don-vip]

In 13678/josm:

see #16188 - rework new test

[Don-vip]

With #16264 I agree: we shouldn't raise a warning when a waterway crosses a residential area. This is quite common.

[Don-vip]

In 13721/josm:

fix #16264, see #16188, see #16189 - reduce validator false positives:

• increase tolerance of buildings with almost right angle to 1 degree
• detect crossing residential areas only with themselves and buildings
• fix crossing railways messages