Re: [...] my rant on reductions...

[Ivan Godard <igodard@xxxxxxxxxxx>]

Amen!

There have been reasonable and unambiguous proposals in this and many 
other areas where the draft is failing. Unfortunately, critical 
components of the committee have not cared to understand the issues, the 
proposals, or the state of the art outside the committee. The committee 
will find a solution consonant with 21st century specification 
technology, or it will be bypassed by folks like the Open Source 
Numerics groups Dan recently responded to.

Yes, these are important issues. Yes, many of the proposals to date 
address pressing problems of portability, performance, and unambiguous 
implementation. No, there is no solution in Fortran II. No, there is no 
DWIM button.

Sorry, guys.

Ivan

Jason Riedy wrote:

> And r754@xxxxxxxxxxxxxx writes:
> - 	Would you like to put your expertise where your mouth 
> - 	is & join us?
>
> David Bindel and I compared a few possible 
> implementations of the scaled alternate exception 
> handling pretty much exactly three years ago.  We
> discussed some of the options' performance 
> considerations on current platforms and made 
> implementation recommendations.  We had running
> code and timings.  We also included a "scarecrow 
> proposal" recommending a scaled type, which is the 
> right way to solve this problem from (what should 
> be) the 754 level.
>
> Now I'm more interested in fixing the basic model
> so these options can be expressed clearly to 
> implementors.  But that work is in the unfunded
> queue.
>
> Jason
>
>