Re: NRC Public Briefing on Ada

[dewar@merv.cs.nyu.edu (Robert Dewar)]

iBob Love said

"This is pretty absurd.  Simulators and mission planning tools are
getting
closer together all the time.  The simulator uses the same information
and mostly the same hardware, as the real vehicle."


To emphasize this point, one of the really tough problems we faced with
GNAT recently was a customer who was building a simulator and wanted
to use EXACTLY THE SAME CODE, character for character, that was in the
operational vehicle for the simulator.

That's a reasonable demand, although given that different compilers for
different languages (Ada 83 vs Ada 95) were involved, a little tough to
meet.

Turns out we had to add a couple of pragma Pack's and that was all (the VADS
compiler lets a size clause do implicit packing, something that is contrary
to the implementation advice in the Ada 95 RM, and something which NAT
GNAT does not feel like emulating).

Anyway, it took a bit of doing, but we got permission to add the pragma
Pack.. In fact expecting quite this level of compatibility in a situation
like this where the operational and simulation hardware are totally
different is a bit extreme, but the idea of writing the operational 
software in language x and the corresponding simulator software in language
y is totally nuts to me, for any possible choices of x and y!