Geometry: call clearBounds after a move or resize

[fredj]

The geometry bounds change after a move() or resize().

[crschmidt]

fredj: Pierre and I discussed this in #666. We came to the conclusion that clearBounds might have a significant performance impact, and I wanted to get Tim's feedback on whether this was the right way to do things. I'll poke him again and see if he can offer thoughts on that, after which we can either apply this or get a new patch together that does something similar.

Is this to fix an issue other than #666?

[pgiraud]

I think this is somewhat different because it doesn't only deal with VML. If a feature is moved, resized or rotated and the user needs to know its new bounds, the getBounds() method does return the right value because it is "cached".

[crschmidt]

Ah, foolish me: This seems obvious. Hm. But probably needs a test. Can we write one?

[pgiraud]

patch with tests

[crschmidt]

clearBounds is not currently called in the collection either: this seems likely to cause the bug to remain in place. Pierre is looking at testing to see the current behavior.

[pgiraud]

added clearBounds in resize, also simplified tests

[pgiraud]

as said on the IRC:

I was writing a new test on collection move to test bounds update

and surprisingly the test pass well I first couldn't figure out why, and was looking what was wrong in my test but I discovered that clearBounds also clears the geometry's parent bounds

[tschaub]

Thanks for catching this, correcting it, and writing tests.

All pass in IE/FF. (I tweaked the tests a bit - the bounds var was not used any more.)

[tschaub]

in w/ r3998

[pgiraud]

tests miss getBounds() call so have no interest

[pgiraud]

I agree with Tim to say that the bounds var is not use any more. Though the getBounds() call is mandatory.

[tschaub]

Thanks again - I apologize for not fully understanding this before making edits.