Re: IGS client server protocol.

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

NNGS and IGS are not 100% compatible in their messages, so you actually
have to reverse engineer both is you want a fully general client.  The
IGS source code that NNGS used is just the header file I mentioned in
my last message.

When you write your client be aware that there is some trickiness involving
waiting for prompts before sending new information.  There are some races
that can cause your sent data to be lost.

David

At 05:30 PM 10/5/99 +0200, Jan Ramon wrote:
>On Tue, 5 Oct 1999 rbrown@xxxxxxxxxxxxxxxxx wrote:
>
>> There are no specs.  As far as I understand it, all the client writers
had to
>> use trial-and-error to discover for themselves the behavior of the server.
>
>I agree to some extend, but i think there are some documentation and hints
>available:
>- nngs source code is available.  there have been flames on rgg which
>  contained so much information that i could derive that some people
>  believe that the nngs and the igs source codes have some common
>  ancestor (and others dont believe this). For persons who think there is 
>  some chance that the first people are correct, this information could
>  reduce the amount of error (and trial).
>- some time ago there was a system broadcast message one igs "client
>  writers, contact Tim Somename" (but i dont remember the name).  Perhaps
>  finding and contacting this person would help or at least give some
>  better guarantee that help can't be expected.
>- for a client which makes use of some "client" features of igs, 
>  see http://w3c.teaser.fr/~jlgailly/go.html
>
>jan
>
>
>
>
>