Re: Progress - OpenGo...

[Jeffrey Greenberg <jgberg@xxxxxxxx>]

I agree with this.
If GMP were an 'interface', then we could just change the transport layer, but it
really is a wire-protocol.  I am not aware of a tcp/ip definition that Markus is
referring to.

Also, if you want to test more than one engine at a time, say a pool of engines,
then GMP becomes truly unworkable if you really have to connect up all those
modems.  A networked scheme would be much more convenient.

It may be that the IGS or NNGS protocol is sufficient... If so, then we don't
have to invent another standard, just encapsulate the existing one into an
interface... or has that already been done?  OTOH, IGS or NNGS may not be 'free'.

jeff


Antti Huima wrote:

> On Thu, 13 May 1999, Markus Enzenberger wrote:
>
> >
> > On Wed, May 12, 1999 at 11:27:05PM -0700, Jeffrey Greenberg wrote:
> >
> > > 4. how about a better test interface than the Go Modem Protocol, based on
> > > TCP/IP?
> >
> > Just curious: is there anything wrong with using Go Modem Protocol
> > on TCP/IP?
> >
> > - Markus
>
> IMHO, Go Modem Protocol is horrible. It could have been suitable for modem
> communications many years ago, but now that TCP/IP "rules", the parity
> bits and the amount of compression used is obsolete, I think. I think we
> should develop a new protocol for 2000's.
>
> --
> Antti Huima
> SSH Communications Security Oy

begin:vcard 
n:Greenberg;Jeffrey Greenberg
tel;fax:650-325-0488
x-mozilla-html:TRUE
url:http://www.concentric.net/~jgberg
adr:;;;;;;
version:2.1
email;internet:jgberg@xxxxxxxxxxxxxxxxx
x-mozilla-cpt:;23728
fn:Jeffrey Greenberg
end:vcard