[patch] Area style on outer way not flagged when different river area taggings are combined

[zelonewolf@…]

This issue is present as of JOSM version 17653.
I am attaching a simple test case to demonstrate the issue.

When an "area" multipolygon contains a closed way outer member that is also an area, the validator should flag the warning "Area style on outer way". However, this does not seem to be occurring under the following combinations:
1) The multipolygon is tagged waterway=riverbank and the member is tagged natural=water+water=river
2) The multipolygon is tagged natural=water+water=river and the member is tagged waterway=riverbank

The test case shows That the validator completely ignores these incorrect cases while correctly detecting issues in cases where both the outer member and parent multipolygon are both using the same river area style.

[zelonewolf@…]

Simple test case demonstrating the issue

[GerdP]

The code in MultipolygonTest checks if the mappaint style on the outer way is similar to that of the multipolygon. If yes, the informational message With the currently used mappaint style the style for outer way mismatches the area style is produced, if not, the warning Area style on outer way is produced.
This explains the special behaviour.
It's probably better to always show the warning Area style on outer way and simply remove the code for the other message.

[GerdP]

always show warning when outer way has area style, adapt unit test data

[GerdP]

Hm, with this patch JOSM sometimes produces two warnings for a multipolygon where the outer way has the same tag as the relation. Tried the test file for #20473 and got

Area style on outer way (1)
Multipolygon outer way repeats major tag of relation - Same tag:'leisure'='water_park' (1)

[Don-vip]

Milestone renamed

[GerdP]

Not sure what to do here. Should I remove the test from TagChecker which creates the 2nd message?
Are there really no cases where a closed outer way in a multipolygon is allowed to have an "area style"? If yes there is no need for the detailed check in TagChecker and I can revert r17501.

[GerdP]

reverts also the changes from r17501

[skyper]

The warning about style is an internal warning. Yes, as I can use areas in multiple MP-Relations as outer and still define the closed way with more specific tags it is possible that the style in use mismatches the area style. I think the current warnings are sometimes too strict.

[GerdP]

Hm, maybe I should stick to the first patch.
The test in TagChecker only complains when the tags are equal, either with
Multipolygon outer way repeats major tag of relation or with
Multipolygon outer way repeats tag of relation.
So, it is about a different issue.
In what situation do you see too strict warnings? With different mappaint styles?
I could use the code in TagChecker to produce a better message. E.g. instead of Area style on outer way we may report
in group Multipolygon outer way has area tag the message area tag on outer way:''{0}''=''{1}''

[skyper]

example with MP mismatching area

[skyper]

+1, for better warnings

Think we have a discussion about Mappaint style warning (rewording) on another ticket, too. Actually, I do not use other styles defining areas, so I have never tested without default style.

The strictness might be ok, but why can I silence this warning by using MPs? In my example​ I get the warning but if I change the outer way into another MP I get no warning. Is this intended?

Another question is, if we have these mappaint style warnings as own category or if we use the style in other warnings and silence one in favor of the other. I prefer a cleaner cut and have other validator warnings not rely on the style.

Coming back to this report, we have a special case with two tagging systems (riverbank <=> water) which might need special treatment.

[Don-vip]

Milestone renamed