Hi all,
during the last days I've tested the branch.
See also
http://gis.19327.n5.nabble.com/bounds-branch-tp5813741p5813749.html
I found that the changes in the img files are not important, they happen
because parts of a multipolygon relation are saved in different orders,
so no change should be visible.
Reg. the unpredictable(?) differences in run times:
I am still able to reproduce this in Eclipse, but I found no explanation.
From time to time CPU intensive routines are executed faster (e.g. ~40 secs instead
of ~60)
If I hear no complains I'll merge the branch to trunk on sunday.
Gerd