Complex Division Patch (reprise)

[craig@jcb-sc.com]

>The proposed complex division changes violate both IEEE and ISO.
>Please read the standards documents.
>Please do not make these changes.

Perhaps it would help if you'd identify the documents in question,
perhaps citing the relevant sections and/or providing URLs for
those of us who don't have (and perhaps cannot afford) hardcopies
of these standards?

Also, are you saying they violate the standards *regardless* of the
input language?  AFAIK, g77 already uses the complex-division
algorithm Toon's patch now open-codes.  Are you saying this violates
the IEEE and ISO standards pertaining to Fortran, i.e. that g77
already violates these standards, even *without* Toon's patches
(egcs 1.1.2, for example)?

Nobody has commented on my earlier proposal to make Toon's patch
conditional on what the front end wants, rather than -ffast-math.
I made it about a week ago.

It'll be too bad if people keep sniping at Toon's work, rather
than carefully reading the comments about it and carefully identifying
their concerns, resulting in the work not making it into 1.2,
especially since the performance benefits for g77-compiled code
could be substantial for some types of code.

        tq vm, (burley)