On Tue, Dec 11, 2001 at 10:34:01AM +0100, Lars Gullik Bjønnes wrote: > | It's an optimization that admittedly never got profiled. But note that > | the copy-construction involves deep copies of the nucleus and those are not > | cheap... > > OTOH wouldn't self assignment almost be considered a bug? Or is it > something that we must expect to happen in mathed?
I guess it is pretty unusual. Since MathAtoms with a few exceptions live within a MathArrays self-assignments should be caught at a higher level by std::vector. Maybe I should add a message there just to find out. Andre' PS: Can I apply the eufrak stuff? -- André Pönitz .............................................. [EMAIL PROTECTED]