--On Tuesday, March 9, 2004 11:17 PM +0900 Yasuki Arasaki
<[log in to unmask]> wrote:
> Well, I was thinking of just skipping integers and dealing with
> only real types, but..
Yes, I could see that. I was saying that I think that dealing with
integers is at least as important. If you think that sticking _whatever
on the end of every literal real is a bother (and I agree with you),
wait until you have to do it for every integer literal!
> Maybe a question that has to be asked first
> is if the defaults will change in the future?
>
> By the time F2003 is as widely adopted as F95 is today, 64-bit
> integers will be the more natural one on most hardware (that's PCs).
> Will Fortran have default 64-bit integers then? If it will, then
> precision of default real will be made better just by waiting,
> and it may be the best solution.
>
> Maybe it's more productive to think of ways to ensure that the
> de-facto default change to 64-bits?
I once thought this was the answer to this quandry - that the problem
would largely solve itself because the default would eventually become
64 bits, which would cover the most common needs. I am no longer
convinced of that. There is a large code base that assumes 32-bit
single precision. This code base is growing, not shrinking, because
even though I try to write code that avoids such assumptions, I'm
clearly in a minority.
How to accommodate the existing code and user base, and also to
simplify writing new code, and to allow this all to work together,
with old and new code being maintained and mixed in the same application...
that's a lot harder than the problem of starting fresh.
I have become of the opinion that the defacto default is likely to
stay at 32 bits, even as larger sizes become widely available. But I'm
not nearly as certain of this opinion as I once was of my opposite
opinion. :-) I know that I'm far from the first to have noticed that
the more experienced I get, the less knowlegable I seem to become about
some things. Ah for my teen years when I knew everything. :-)
|