Rotated elements jump when repeatedly selected

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Draw a text element and another random element
2. Select both of them
3. Group them
4. Rotate the group, It does not seem to happen if the rotation amount is 
divisible by 90. With the angle 83.35472106933594 it looks like it's moving 
around in a spiral.
5. Ungroup them
6. Select the text. Notice that it moves slightly
7. Repeat step 6 a few times and the amount it jumps increases

What is the expected output? What do you see instead?
It should stay still, but it jumps around.

Taking note of the svg transform values, the rotation angle remains 
consistent, but the center of rotation changes.

Please use labels and text to provide additional information.

Original issue reported on code.google.com by Antimatter15 on 9 Feb 2010 at 5:46

[GoogleCodeExporter]

Hm, not seeing this yet, though it sounds familiar (from the past while working 
on
transforms). Can you clarify the spiral comment in #4.

Original comment by codedr...@gmail.com on 9 Feb 2010 at 9:39

[GoogleCodeExporter]

1. Draw a Rectangle and a text element.
2. Select both of them by drawing the rubber-band
3. Click the group button
4. Rotate the entire group by 83.35472106933594 (some other rotations might 
work too)
5. Ungroup the group
6. Select both of them by drawing the rubber-band selector thing
7. Repeat step 6 about 20 times, the movement isn't noticiable for the first 5 
or so 
times but then it gets more pronounced.

Original comment by Antimatter15 on 9 Feb 2010 at 10:07

[GoogleCodeExporter]

Wow - that is downright spooky! :)

At about the 10th rubber-band selection I was thinking you were pulling my leg, 
then
it started to become detectable a few selections later.

As you mentioned, this is because the rotational center gets incrementally
off-centered from the bbox the more times you select it.  Must have to do with
recalculateDimensions().

Original comment by codedr...@gmail.com on 9 Feb 2010 at 10:16

• Added labels: NeededFor-2.5

[GoogleCodeExporter]

Note that this happens for *ANY* rotated element and the elements do not have 
to have
been grouped and then ungrouped.  You just have to do it enough times (like 30 
or so
in some cases).

A workaround seems to be dragging the element somewhere (wherein the rotational
center is back to being the center of the bbox).

Wow, good catch!

Original comment by codedr...@gmail.com on 9 Feb 2010 at 10:21

• Changed title: Rotated elements jump when repeatedly selected

[GoogleCodeExporter]

Is this bug serious enough that it needs to be fixed in 2.5? codedread, think 
you can
fix it? Looks like it'd fall within your realm... :)

Original comment by adeve...@gmail.com on 30 Mar 2010 at 11:52

[GoogleCodeExporter]

You notice that codedread-shaped hole in the door?  That's me running away 
screaming...

In all seriousness, I don't think I can look at this in the short term, but 
maybe in
a couple weeks.

Personally I can live with this in 2.5, as it doesn't seem to happen unless you
repeatedly select/de-select.  Granted it's definitely something that needs to be
fixed though.

Original comment by codedr...@gmail.com on 30 Mar 2010 at 11:59

[GoogleCodeExporter]

Original comment by codedr...@gmail.com on 14 Apr 2010 at 7:43

• Added labels: NeededFor-2.6
• Removed labels: NeededFor-2.5

[GoogleCodeExporter]

Issue 550 has been merged into this issue.

Original comment by adeve...@gmail.com on 3 May 2010 at 12:48

[GoogleCodeExporter]

Since it also occurs when repeatedly moving a rotated element I thought I'd try 
fixing 
it...r1572 seems to take care of it. Need some good testing to make sure I 
didn't 
break anything, though.

Original comment by adeve...@gmail.com on 18 May 2010 at 2:55

• Changed state: Fixed
• Added labels: NeededFor-2.5
• Removed labels: NeededFor-2.6

[GoogleCodeExporter]

i'll put it through the ringer

Original comment by dezigns...@gmail.com on 18 May 2010 at 5:02

[GoogleCodeExporter]

looks good so far but I found a new issue.

Original comment by dezigns...@gmail.com on 19 May 2010 at 5:42

[GoogleCodeExporter]

Okay, thanks...could you file the new issue? :)

Original comment by adeve...@gmail.com on 19 May 2010 at 12:30