FW: First implementation of Multi-source library in SSM networks announcement

Ivan R. Judson judson at mcs.anl.gov
Mon Sep 8 11:32:42 CDT 2003



> I thought this was in the realm of the capability stuff 
> that's already 
> there; I assume that that's why the IP address solution is trivial.

That may be true, that's a Tom bit of information.
 
> It would be good to be able to enable research into some of 
> this SSM stuff 
> without having to go through a round of design discussion, 
> especially if 
> it's trivial in the current framework. 

Yes, I agree, but this is exactly the kind of change the causes disruption,
bugs, and support problems when we do it in the "development" code (which is
in use by too many people NOT developing or doing research). That's the
whole point of us having a design cycle and the value of having a
development effort, to reduce end user difficulty, make the code
maintainable, and to spread the effort among the entire development team. 

Therefore, I don't agree that we should make this minor change without
design discussions, unless it can be shown that it has no effect on existing
users. The people interested in researching SSM should feel free to request
this feature, and consider whether or not our response is soon enough (we'll
try to make it soon enough, but we have other constraints). If it's not soon
enough, they might want to make this change in a cvs copy and let us know
how it works and feed the mods back into the dev tree via ag-tech.

Straying from this development model is likely to cause us all sorts of
grief and I don't think we need to court grief.

--Ivan

--Ivan 




More information about the ag-dev mailing list