There are still some inconsistencies between IntPoint/FloatPoint, RoundedRect using floats, etc.
Why don't we use template classes for Point/Rect/Size?
For one, there are various overloads and implicit conversions that make sense to do from int->float but not the other way around. Functions like isZero would also need specialization for floats.
Created attachment 101405 [details] Patch
Comment on attachment 101405 [details] Patch I think this is good. It might be worth investigating sharing more code between IntPoint/FloatPoint, but at a glance it doesn't seem trivial. We could handle a lot of cases with partial specialization.
Comment on attachment 101405 [details] Patch Thanks for the review!
Comment on attachment 101405 [details] Patch Clearing flags on attachment: 101405 Committed r91417: <http://trac.webkit.org/changeset/91417>
All reviewed patches have been landed. Closing bug.