ian@oravax.UUCP (Ian Sutherland) (03/05/90)
In article <8231@hubcap.clemson.edu> billwolf%hazel.cs.clemson.edu@hubcap.clemson.edu writes: > And of course, other languages do not enforce portability by having [...] > a compiler validation > process which ensures that Ada compilers all perform according to > exactly the same specifications (again ensuring portability). Are you referring to the fact that a validated Ada compiler must execute a certain collection of programs (the validation suite) in a certain way? If so, this hardly establishs anything as sweeping as "Ada compilers all perform according to exactly the same specifications", unless you've got a pretty weak notion of the term "specification". This claim is especially suspect in view of the number of things about Ada execution which are left unspecified by the ALRM (e.g. parameter-passing mechanisms). An Ada compiler that uses one mechanism will, in certain circumstances, perform quite differently from one which uses another. The ALRM's answer to this (which essentially amounts to "don't write programs that notice the difference") seems somewhat less than satisfactory to me. Don't get me wrong. I believe that the measures taken to enhance the portability of Ada do enhance it. I just think the claim you made above is a bit extravagant. -- Ian Sutherland ian%oravax.uucp@cu-arpa.cs.cornell.edu Sans Peur