False positive conflict detection with reverter plugin

[skyper]

Thought there was already a ticket about it, but I did not find it.

I get these false positives when reverting but I am not sure if it does happen with other merges,too.
The problem appears with younger nodes than the reverted changeset.

1. revert changeset 14914753.
2. got 64 conflicts
3. look at conflict of node id:980535480
4. there is no conflict and you do not have any option to change anything.

The reported conflict is useless.

I am not sure weather these are conflicts and you need some options in the conflict manager or not and the whole conflict should be dropped.

[Don-vip]

It also produces an exception if we click on the left button of the tags tab:

Identification: JOSM/1.5 (5929 SVN en_GB) Windows 8 64-Bit
Memory Usage: 194 MB / 1813 MB (64 MB allocated, but free)
Java version: 1.6.0_45, Sun Microsystems Inc., Java HotSpot(TM) 64-Bit Server VM
VM arguments: [-Dfile.encoding=UTF-8]
Dataset consistency test: No problems found

Plugin: reverter (29559)

java.lang.IndexOutOfBoundsException: Index: 0, Size: 0
	at java.util.ArrayList.RangeCheck(ArrayList.java:547)
	at java.util.ArrayList.get(ArrayList.java:322)
	at org.openstreetmap.josm.gui.conflict.pair.tags.TagMergeModel.rememberDecision(TagMergeModel.java:136)
	at org.openstreetmap.josm.gui.conflict.pair.tags.TagMergeModel.decide(TagMergeModel.java:164)
	at org.openstreetmap.josm.gui.conflict.pair.tags.TagMerger$KeepMineAction.actionPerformed(TagMerger.java:283)
	at javax.swing.AbstractButton.fireActionPerformed(AbstractButton.java:1995)
        ...

[Don-vip]

Ah, the exception is a dupe of #5680.

[Don-vip]

Fixed in ​[o29561].

[Don-vip]

Looks like it is not yet fully fixed as I found several useless conflicts while reverting changeset 15875597.

[Don-vip]

(with ways, I think I have fixed the problem only for nodes in fact)

[GerdP]

I can still reproduce useless conflicts when loading the attached file and reverting 15875597 with current binaries,
e.g. way 75022047 is not modified but I see a conflict. I'll have a closer look again.

[GerdP]

I fear I don't fully understand the concept of the conflicts produced by reverter.
Example:
I download a small area around node 5145791503 which - as of now - has a history of 3 versions. The data
also contains the way 529734514.
Let's assume I like the version 1 better, so I revert the cs 52802997 which changed it to v2.
I revert with "revert selection only" and I get a conflict (unexpected).
The code in the plugin detects a conflict because it compares v2 with the version in my data (v3)
and finds that the coordinates are different. This is the reason that the conflict that is created.
The conflict displayed in the dialog also claims that version 3 of the node is not member of way 529734514,
which is simply wrong as the node is still a member of the way (on the server and also in my data).
Am I right that my revert should not produce a conflict?

[GerdP]

Please review: This patch suppresses useless conflicts which are typical when you revert a changeset that was already reverted by a more recent cs