Do not show warning about modified incomplete multipolygon when only tags were added/modified

[skyper]

What steps will reproduce the problem?

1. Have an incomplete multipolygon relation
2. add a tag to the relation
3. run validator

What is the expected result?

No warning

What happens instead?

Incomplete multipolygon relation was modified (1)

Please provide any additional information below. Attach a screenshot if possible.

Please, only show this warning if changes in members or their order were made but not if only tags of the relation are added/modified. Thanks

Relative:URL: ^/trunk
Repository:UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b
Last:Changed Date: 2020-01-23 22:57:46 +0100 (Thu, 23 Jan 2020)
Revision:15755
Build-Date:2020-01-24 02:30:53
URL:https://josm.openstreetmap.de/svn/trunk

[pangoSE]

I am also affected by this.

[GerdP]

This is intended, see #18240. With an incomplete MP you don't even know what area is covered and whether its geometry is correct.

[skyper]

Replying to GerdP:

This is intended, see #18240. With an incomplete MP you don't even know what area is covered and whether its geometry is correct.

Oh, ticket is still open, better comment there ?

I did not change the geometry but just tags, therefor this is always a false possitive.

[GerdP]

Why is it a false positive?

[skyper]

Replying to GerdP:

Why is it a false positive?

In my understanding, this warning should only be raised if the geometry was modified. Modifying tags should always treated the same not depending on the object type.

[GerdP]

Unfortunately it is very difficult to detect a change in geometry.
Changing the message to something like "Incomplete relation was modified" is probably also not what you want.
The warning is meant to alarm users that they are changing objects without even knowing the geometry.
OTOH this may also triggered when you split a way at an existing node. This action only changes the number of members but not the geometry or the tags. When I added the code I thought that one should never modify a relation without verifying it. The result could be multiple incorrect versions.
Question is: Remove the message or tell users to add it to the ignore list once they understood what it means?

[skyper]

Replying to GerdP:

Unfortunately it is very difficult to detect a change in geometry.

I know.

Changing the message to something like "Incomplete relation was modified" is probably also not what you want.

"Incomplete multipolygon relation was modified" is better, later "Incomplete multipolygon/boundary relation was modified

The warning is meant to alarm users that they are changing objects without even knowing the geometry.

Think, this is a too general approach as there are many situations where the changes are harmless.

Question is: Remove the message or tell users to add it to the ignore list once they understood what it means?

Ignore list is not good as you will never get this message for the relation again, no whether what you change.

The better JOSM gets in automatically handling critical situations and otherwise directly warning about it, the fewer this warning is needed.
How about showing a warning directly when an action is changing the relation in a critical way like `Incomplete <type> relation was modified. Valdidator needs all members to be downloaded in order to run test. Download (incomplete) members now ?´ and offer to download all (incomplete) members with option to remember answer for action, session and always like adding a duplicated member to a relation in relation editor.
Deleting a member and combining ways (normal mode) already show warnings but need an additional one, splitting ways is on the way and not critical, so we have problems when changing child nodes or roles of member ways and adding members without any warning left as critical for multipolygon/boundary relations:

• deleting end node
• unglue end node
• extending way (new end node)
• moving child node of way with membership (self-intersecting)
• adding new (unconnected) member + role
• changing role of a member

Meanwhile, I can ignore the warning and so far only two persons did argue against it in certain situations.

[pangoSE]

Replying to skyper:

Replying to GerdP:

Unfortunately it is very difficult to detect a change in geometry.

I know.

Changing the message to something like "Incomplete relation was modified" is probably also not what you want.

"Incomplete multipolygon relation was modified" is better, later "Incomplete multipolygon/boundary relation was modified

The warning is meant to alarm users that they are changing objects without even knowing the geometry.

Think, this is a too general approach as there are many situations where the changes are harmless.

Question is: Remove the message or tell users to add it to the ignore list once they understood what it means?

Ignore list is not good as you will never get this message for the relation again, no whether what you change.

The better JOSM gets in automatically handling critical situations and otherwise directly warning about it, the fewer this warning is needed.
How about showing a warning directly when an action is changing the relation in a critical way like `Incomplete <type> relation was modified. Valdidator needs all members to be downloaded in order to run test. Download (incomplete) members now ?´ and offer to download all (incomplete) members with option to remember answer for action, session and always like adding a duplicated member to a relation in relation editor.
Deleting a member and combining ways (normal mode) already show warnings but need an additional one, splitting ways is on the way and not critical, so we have problems when changing child nodes or roles of member ways and adding members left without any warning left as critical for multipolygon/boundary relations:

• deleting end node
• unglue end node
• extending way (new end node)
• moving child node of way with membership (self-intersecting)
• adding new (unconnected) member + role
• changing role of a member

Meanwhile, I can ignore the warning and so far only two persons did argue against it in certain situations.

I agree this warning should be improved and I like your suggestions skyper.

I recommend adding a dialog after download/in download to download all members of relations intersecting the area. This would mitigate the need for this warning completely. If the user has a large multipolygon intersecting (with >20,000 nodes) we should instead show a warning about the size and complexity and warn/recommend splitting up the polygon.