Re: Ada Core Technologies and Ada95 Standards

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

Ken asks

"If the writers of the ACVC tests are expected to write tests that reflect how
Ada is really used, how do they gain this information? For that matter, when
looking at a particular test, how do I judge whether that test reflects common
Ada usage? I know how I use Ada, but how do I know whether my usage is common
or "marginal"?
"

First of all, you are way ahead just asking the question "How would this
feature be used in a real program?" That question was not even asked
in the test generation philosophy of the earlier ACVC suites.

Very often, the answer is pretty obvious and non-controversial.

But it is true that sometimes it is not so obvious. The process we
follow is that tests are reviewed by a review team that is coposed
of experienced users, implementors, and testers. If all members of
the review team agree that a given test is realistically usage-oriented,
that is certainly not a guarantee that this is the case, but it is
a useful indicator.

There is no other way to do things *at this stage*, since there is not
enough experience with many of these features to actually measure usage
patterns, though if there is an ACVC 2.2 to follow on, then you could
certainly consider such measurement.

As I mentioned in an earlier message, our anecdotal experience with GNAT
is encouraging. IN broad terms, we see the new Ada 95 tests have similar
general testing profiles to the tests that come from our users code (and
which presmably are definitely user-oriented).

As for your last question, all I can say is that is why we have more
than one person on the review team. It helps avoid individual
idiosyncratic usage influencing the suite (note I said help, not
avoid, this is definitely not a purely objective process).

But as I said before, why not look at the 2.1 tests and see what you
think in comparison to the ACVC 1.11 tests. They are availabe for
your perusal, and comments are welcome!