On 14/04/2012 1:02 AM, Paul Gilmartin wrote:
On Sat, 14 Apr 2012 00:44:19 +0800, David Crayford wrote:
What I find the most disappoinging about that list is it forces you to
FLOAT(IEEE)! How useful is that for most assembler programs? I suppose
it's to keep the size of the runtime down
to only support functions for one floating point standard. That really
sucks! I could care less about ARCH(0)
Look forward, not back. HFP is _so_ 20th Century. And how much
floating point of a particular flavor is needed in the kernel-type
code which is the target environment of METAL-C?
Disabling DFP may incur greater burdens.
The line of business I'm in requires HFP. I see no reason why the
codegen can't do both. It's an unnecessary constraint.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
