Re: Ada Core Technologies and Ada95 Standards

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

Wes said

"Point Three:
I think KG has missed an important point that RKBD has failed to mention--
the DoD's goal in mandating validation has nothing to do with quality.
Validation assures the DoD that they are using "ONE language" instead of
123* very similar languages."

That's a really good point, and is indeed historically a strong part of the
reason (maybe really *the* reason) that validation is required by DoD.
Now that Ada is no longer effectively trademarked (actually it probably
never was a valid trademark according to some, but anyway certainly is
not now), someone could call something Ada which was VERY far from being
full, or accurate, Ada. Certainly subsets and dialects could proliferate
under this name. Validation, at least within the DoD, effectively prevents
this from happening. Sure we debate occasionally whether some minor feature
is or is not optional, or whether some minor addition is or is not an
extension, but these discussions are very much at the edge of the language,
and validation has indeed stopped dialects and subsets appearing.

Note that the early stages of Ada 95 deliberately *do* allow subsetting
in the transitional period, but that ends next year, and in 1997, once
again, usable Ada compilers must compile the entire language.

Well I guess we have to address the special needs annexes when I say this.
The point with these annexes is that they are basically codified optional
library packages. Of course one has always been able to add library
packages to an implementation. The SN annexes merely codify how certain
library package functionality should be added.

At the moment, only GNAT implements all the annexes, but I think we will
find that eventually all Ada 95 compilers will implement all the annexes,
with the possible exeption of distribution, since this is LARGE part of
the language with significant compiler consequences. Also of course we
do not expect specialized compilers like the Patriot-2 compiler to implement
informatoin systems stuff, or for the Ada-Java implementation to implement
systems programming stuff, but for general purpose compilers, I would expect
pretty complee implementation of the annexes eventually.