Re: [boost] Re: Standardization of Boost libraries

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-03-07 10:09:43

----- Original Message -----
From: "Toon Knapen" <toon.knapen_at_[hidden]>

> > Alas, until we can solve some of the low-level issues with
> > initialization and aliasing, there will still be efficiency
arguments
> > for Fortran.
>
> If the next standard also adopts the C99 restrict, I guess we solved
the
> aliasing issue. Or am I missing soth.

I have to admit that I'm not familiar with restrict at a deep level, but
I'm under the impression that it doesn't solve a lot of the problems
that we really want to solve. Jeremy Siek, Andy Lumsdaine and I had a
pretty intense discussion about this about 8 months ago, but I think
I've forgotten most of that. Anyway, I think these guys had figured out
that we need something really complicated ;^{

-D

> What's the problem regarding initialisation (not taking into account
the
> std::complex as being discussed in the other thread) ?

Just that, AFAIK, but generalized to cover types other than complex<>.
There may be other issues, but I think aliasing is the big one.