| gary.l.scott@lmco.com 2005-02-24, 3:59 pm |
|
Dan Nagle wrote:
> Hello,
>
> On Thu, 24 Feb 2005 03:02:15 GMT, "James Giles"
> <jamesgiles@worldnet.att.net> wrote:
>
>
> <snip requoted>
>
that, just[color=darkred]
nothing[color=darkred]
>
> The TYPELESS proposal solves several problems:
>
> it standardizes and rationalizes existing behavior
> of many compilers regarding BOZ constants,
>
> it is exactly what certain applications want
>
> it doesn't introduce a new class of address (saying "all
> you must do is reckon the bit-per-word part" is saying
> it's a new class of address-- all you have to do to convert
> between character addresses and word addresses is reckon
> the bytes-per-word part)
>
> because TYPELESS uses existing addresses, it may be expected
> to be very efficient
>
> IMHO, the first reason is enough to standardize it.
>
I'm not sure I have a problem with standardizing BOZ constant behaviour
relative to integer data types (I didn't actually know there was a
problem there). But, this is definitely not a substitute for "bit
string" data type. A bit string might require 256 bits or 256121 bits.
It depends on what type of externally defined data type you're trying
to emulate/match.
> --
> Cheers!
>
> Dan Nagle
> Purple Sage Computing Solutions, Inc.
|