Re: What's stopping you from using Ada for your next commercial project?

[Dan <dan@irvine.com>]

On Mar 11, 2:40 pm, "Randy Brukardt" <ra...@rrsoftware.com> wrote:
> "Peter C. Chapin" <PCha...@vtc.vsc.edu> wrote in messagenews:alpine.WNT.2.00.1103110703470.1228@WHIRLWIND...
>
> > On Fri, 11 Mar 2011, Manfred Kremer wrote:
>
> >> Wonder what happened to the Ada Compiler Validation Suite.
> >> Was this abandoned or is it still maintained to reflect recent changes to
> >> the ARM?
>
> > Didn't it become ACATS?
>
> >    http://www.ada-auth.org/acats.html
>
> > If so, it has been updated to follow the latest standard.
>
> Right, but keep in mind that the purpose of the ACATS is to ensure
> commonality between Ada implementations, not to find bugs per-se. Of course,
> it does have the effect of finding bugs as well.
>
> In particular, the ACATS does not have a lot of tests that combine various
> features together. That's simply because the emphasis is on ensuring that
> every feature is there and works as expected -- bugs are much more likely to
> surface when features are combined (as in the OPs question).
>
> So while the ACATS raises the quality of Ada compilers, it is not going to
> come close to eliminating bugs in those compilers.
>
>                                            Randy.

When the ACVC was originally developed, the intent was that it *would*
come close to eliminating bugs in compilers.  It was certainly
recognized that a finite number of tests developed without knowledge
of actual compiler behavior couldn't detect all bugs.  But it was
intended that the ACVC would be continually enhanced, so when actual
compiler bugs did surface, the ACVC test objectives and corresponding
tests would be reexamined and strengthened if necessary.   Validation
certificates expired annually, so such compiler bugs would be
eliminated in the next round of testing.

This is explained in John Goodenough's 1981 paper at:
   http://www.computer.org/portal/web/csdl/doi/10.1109/C-M.1981.220496
on page 59:

"The IG will be a living document. As implementers ask
questions about the standard, as implementation subtleties
are discovered, and as errors in compilers turn up,
the IG will be updated by those responsible for maintaining
the ACVC. We expect the IG to minimize unwitting
deviations from the Ada standard, despite the plethora
of Ada implementations likely to emerge in the coming
years. At the very least, Ada implementations should be
closer to standard than those of other widely implemented
"standard" languages."