Visually deprecate old multipolygon styles

[stoecker]

Currently we still support old styles of multipolygon see ElemStyles.java​. I think it is time to somehow deprecate them more than we do now.

• So probably first is to disable the VERY old method of inner=outer completely.
• second is to never display outer styles for complete multipolygon anymore.
• Third is probably to visually mark the no-style on multipolygons as bad for these cases, where the outers have one common style.

Probably instead of removing code, we can turn it of with a default-false variable?

[stoecker]

According to trunk/data_nodist/multipolygon.osm type 1 (very old) is anyway broken :-)

[stoecker]

Cases 07/01 and 06/03 don't even create a warning ATM. That's broken.

[stoecker]

Unclosed areas (multipolygon or not) should no longer be displayed as proper area. They should be marked somehow.

With proper old-style handling cases 03 and 04 of example should be identical. They are not, so this is also broken ATM.

There should be a real cleanup of area style handling.

[Don-vip]

You may add unit tests that use this file for non-regression :)

[bastiK]

Replying to stoecker:

Currently we still support old styles of multipolygon see ElementStyles.java​. I think it is time to somehow deprecate them more than we do now.

• So probably first is to disable the VERY old method of inner=outer completely.

I think we should remove the very old style completely. This should get rid of some ugly special handling in the code.

• second is to never display outer styles for complete multipolygon anymore.
• Third is probably to visually mark the no-style on multipolygons as bad for these cases, where the outers have one common style.

There are basically two styles in use:

(a) area tags on multipolygon relation
(b) area tags on outer way(s)

If in (b), the tags of the outer ways do not match, this is an error.

The 3rd suggestion I understand that you like to discourage the style (b) somehow. Before we can remove support for this, we should get some kind statistics, how much style (a) and (b) is used overall in the data base and maybe how much it is used in multipolygons that have been created / modified in the last year.

What do you mean by suggestion 2? Surely the outer way can be a highway, so a line style is okay.

[bastiK]

Replying to stoecker:

Unclosed areas (multipolygon or not) should no longer be displayed as proper area. They should be marked somehow.

With proper old-style handling cases 03 and 04 of example should be identical. They are not, so this is also broken ATM.

Aren't they identical? Just the tag is different, so there is a another color.

[stoecker]

Replying to bastiK:

Replying to stoecker:

Currently we still support old styles of multipolygon see ElementStyles.java​. I think it is time to somehow deprecate them more than we do now.

• So probably first is to disable the VERY old method of inner=outer completely.

I think we should remove the very old style completely. This should get rid of some ugly special handling in the code.

Fine with me.

• second is to never display outer styles for complete multipolygon anymore.
• Third is probably to visually mark the no-style on multipolygons as bad for these cases, where the outers have one common style.

There are basically two styles in use:

(a) area tags on multipolygon relation
(b) area tags on outer way(s)

If in (b), the tags of the outer ways do not match, this is an error.

The 3rd suggestion I understand that you like to discourage the style (b) somehow. Before we can remove support for this, we should get some kind statistics, how much style (a) and (b) is used overall in the data base and maybe how much it is used in multipolygons that have been created / modified in the last year.

What do you mean by suggestion 2? Surely the outer way can be a highway, so a line style is okay.

Broken thinking. :-) 2 and 3 boil down to "Never care for area styles on multipolygon outer members". Currently with a default false variable to be able to turn it back.

I think our task as editor is to get cleaner style, so we should not wait for the DB anymore, but start acting. Compatibility phase was long enough. It's deprecated for years now. Based on a request on dev mailinglist some month ago.

So my suggestions would be:

• cleanup
• fix 2 broken validator cases
• check or fix validator: any area draw styles on outer ways should be warnings
• disable handling of "outer" styles for multipolygons
• disable display of unclosed areas (polygon or unclosed ways)

Last 2 should be announced in start page.

I had a look at multipolygon/boundary now again and JOSM's (or also my) secret (or not so secret) steps to cleanup database resulted in major reduction of ambiguities. E.g. exclave/enclave is completely dead. Special cases dropped. Boundary type Multipolygon is dying. So a new task ...

[stoecker]

Replying to bastiK:

Replying to stoecker:

Unclosed areas (multipolygon or not) should no longer be displayed as proper area. They should be marked somehow.

With proper old-style handling cases 03 and 04 of example should be identical. They are not, so this is also broken ATM.

Aren't they identical? Just the tag is different, so there is a another color.

Right. Why the hell did I use different styles for each row? No idea.

[bastiK]

In 7563/josm:

see #10529 - remove "very old" multipolygon style (area tags on both inner and outer ways)
(actually, this was never supported for MapCSS styles)

[bastiK]

This is all I plan to do on this topic for now... :)

[stoecker]

In 7565/josm:

see #10529 - fix one validator test case for multipolygon

[stoecker]

In 7566/josm:

see #10529 - add another test case for inner/outer

[stoecker]

In 7567/josm:

see #10529 - add parameter to disable outer-way styled multipolygons, defaults still to true

[stoecker]

Added a parameter now. Question is how the default value should be.

I didn't find the relevant code yet where I can tell JOSM, that unclosed areas aren't areas. The value "pretendWayIsClosed" seems to be wrong place. Any suggestions where to search?