selectaction.cycles.multiple.matches stops cycling the path selection & loses pointer rectangle

[A_Pirard]

JOSM 6060; Java 1.7.0_25.

I was finding the third wheel and right-alt a pita until I discovered selectaction.cycles.multiple.matches.
I think it should be true by default, however: who does click an object several times without reason?
And who would complain for discovering that feature rather than complain for not discovering it?

There is a small bug, however.
If a node belonging to a way of an overlap is control-clicked, the selection of the way cycles.
That makes getting the wanted selection very tricky.

Also, while playing with that, the cycling feature disappeared: click working just the plain old way.
I opened another layer to test there and it was cycling all-right.
I came back to the first layer and it was cycling again.

But in the process, the mouse pointer had lost its rectangle on both layers.

I'm afraid I can't reproduce this, though.

I wondered for months what's inside the rectangle of the mouse pointer until I rtfm.
May I suggest to draw the path horizontally in the rectangle?
I'd be happy to be helpful if you send me or tell me where to easily get the pictures to update.

[AlfonZ]

Replying to A_Pirard:

I wondered for months what's inside the rectangle of the mouse pointer until I rtfm.
May I suggest to draw the path horizontally in the rectangle?
I'd be happy to be helpful if you send me or tell me where to easily get the pictures to update.

Images of cursor modifiers can be found at trunk/images/cursor/modifier.

[skyper]

Replying to A_Pirard:

I came back to the first layer and it was cycling again.

But in the process, the mouse pointer had lost its rectangle on both layers.

The mouse pointer problem is a know issue, but we do not have a clue how to reproduce it so far. It did happen at my systems once last week but activating a non data layer and switching back solved the issue for me.