Re: Re: News - Commercial Version of Hercules - Turbo Hercules

On 10 April 2010 08:05, kerravon86 <kerravon86-/[email protected]> wrote:

> In this case, we don't greatly care that IBM isn't
> being forced to document its interfaces. They have
> published their interface of their own free will
> anyway - otherwise their customers can't use it.
>
> There are some things that aren't documented very
> well (DIAG?) but we can apparently live without
> that for the time being.

You weren't around during the glory days of 2002, when Peter Ward (of
Funsoft) and Phil Payne (ex Amdahl guy with some strange axe to grind
against hercules) spent endless fun hours trolling about the origins
of the Hercules implementation of the SIE instruction for zArch. (It
turns out it *was* documented publicly by IBM, but there's some
question about whether that was a mistake on IBM's part.)

Certainly the general instructions are well documented, but there are
control instructions that are not. DIagnose is not a good example, as
it is entirely model dependent, and is not generally used by operating
systems. But SIE is a good one, as are the Service Processor
interfaces, and several others.

And it's interesting that IBM's letter to TurboHercules appears to be
making noises about the Hercules' use of "proprietary" information. I
wonder if that goes back to the 2002 days, or if there's something
newer.

Tony H.