Re: Derived types and allocatable

> Since conformance testing is identical to the test for
> reallocation (the only difference is what you do if
> conformance didn't match), the only overhead if
> both sides do conform is the test.  

However, the point is that when the compiler knows reallocation cannot 
happen - because the colon form is used or, as apparently in the case of 
the Intel compiler, a switch has disabled the functionality - it can 
completely skip the test. Although I haven't checked, my gut feeling 
tells me that conformability of array assignments is not a constraint in 
the standard, and it's therefore up to the programmer to ensure Bad 
Things Don't Happen and not up to the processor.

> I'm aware that
> some implementors claim the overhead is significant.

Walking through two seperate data structures, comparing them for 
conformability - with the performance of that test being influenced by 
the detailed design of the data structure, which however also needs to 
serve other uses - and with the large amount of conditional jumps that 
entails does strike me as a potentially significant overhead if you're 
not moving megabytes every time.

And as noted above for the Intel compiler, apparently competent people 
have concluded that it's worth the overhead (in a different meaning of 
the word, to be sure) of enabling their customers to control this 
themselves. So there.

	Jan