Ada policy

[Robert Dewar]

Matt Austern wrote:

> The other thing, of course, is that people who are working with
> the FSF compiler and the free test suite can't be expected to
> fix regressions that are only visible in a proprietary test suite
> that they don't have access to. 

Yes, of course, that goes without saying. The same thing is true
for other GCC back ends. You can't fix what you can't see. What
needs to happen here (and this has happened frequently in the
past), is that if a change breaks some test in our test suite,
then we either fix it ourselves, or we produce a distributable
test case (often we try to do this in C since the problem so far
is all concerned with back end bugs). We can't expect anyone to
help if we don't do that!

> The basic question, as far as I can tell, is: for someone who
> isn't a dedicated Ada maintainer, how much responsibility do they
> have for keeping Ada working? If everyone is happy for the answer
> to be "not very much", then most of these discussions about process
> can be put aside.

Well running the ACATS tests is actually a very substantial step.
I think that even *requiring* this is something that needs to be
discussed, because we don't want to erect obstacles in the way of
people working on other languages who have no interest in Ada. Yes
there's a balance here, but I would hesitate to require such testing.
Instead I would try to make it as easy as possible, and encourage it
as much as possible.

Furthermore, if we proceed ahead and add the tests that can be
added, which Laurent has volunteered to help with, that will be
a further level of tests