Hi Kent,
Kent Budge writes:
> The only interesting question remaining is whether valarray has failed
> because it is badly designed, or because the underlying concept (of a
> special type with controlled aliasing) is fundamentally flawed. I am
> sufficiently skeptical of my own ability that I won't rule out the
> former, but my feeling is that the latter is (also?) true.
The concept of having types with controlled aliasing seems OK to
me. It is just the idea of trying to make a *single* type (valarray)
the solution for aliasing that is not such a good idea.
There is a recurring theme of people trying to get language/compiler
support for optimizing particular types: complex, valarray, complex,
intervals, complex, etc. The problem is that the amount of work is
large, and the payoff is quite small. If instead we had general
mechanisms for allowing good optimization of user-defined types, then
we would no longer have to bug standards committee's with requests for
special treatment of this type or that type. Instead we could just
write our libraries, provide some hints to the compiler, and get good
optimizations for all our types. The KAI compiler, with its
lightweight object optimization, already gets us half way there. A
general mechanism for aliasing behaviour might get us the rest of the
way.
Cheers,
Jeremy
--------------------- Object Oriented Numerics List --------------------------
* To subscribe/unsubscribe: use the handy web form at
http://oonumerics.org/oon/
* If this doesn't work, please send a note to owner-oon-list@oonumerics.org
This archive was generated by hypermail 2b29 : Wed Feb 20 2002 - 03:20:11 EST