Re: Gnat and TCP/IP

[Samuel Tardieu <sam@inf.enst.fr>]

>>>>> "Matthew" == Matthew Benjamin <dickery@umich.edu> writes:

Matthew> Dear Folks, I'm just becoming familiar with Ada/GNAT, and am
Matthew> trying to determine how feasible it would be as an
Matthew> implementation language for client/server database type
Matthew> applications. . .

It's as feasible as with other languages such as C, since you can
import any C library function using "pragma Import".

But if your entire application is going to be written in Ada, then you
might consider using the Distributed System Annex which enforces
strong typing and version consistency over a distributed
program. Moreover, you do not have to write your own communication
protocol since everything you need is either in the DSA libraries or
generated by the compiler. Exceptions are of course handled correctly
(you may want to check the Annex E of your Reference Manual if you
want more information about this).

Matthew> Is there a standard TCP/IP implementation for Ada and the GNU
Matthew> Ada compiler? You would expect so, but I haven't heard of
Matthew> one.  (The PAL archives one for VAX/Verdix, but...)

AFAIK there is no de-facto standard TCP/IP implementation for Ada95;
an implementation of the DSA for GNAT is already in alpha-test phase
(contact distribution@act-europe.gnat.com if you want more information
on this) and uses its own implementation of the TCP/IP
interface. Unfortunately, this one is not very well suited for general
TCP/IP usage (this was not its goal).

Matthew> Are people just using external C code?

People use internal Ada code :-) The only thing you have to do is
translate the "*.h" C headers into "*.ads" Ada specifications. C uses
no magic, it calls library functions, and Ada can do the same :-)

  Sam
-- 
"La cervelle des petits enfants, ca doit avoir comme un petit gout de noisette"
                                                       Charles Baudelaire