Re: [PATCH] ada/16086: Accept protected subprogram as generic formal default

[Laurent GUERBY <laurent at guerby dot net>]

On Sat, 2008-04-12 at 11:45 +0200, Samuel Tardieu wrote:
> On 12/04, Laurent GUERBY wrote:
> 
> | On Sat, 2008-04-12 at 03:31 +0200, Samuel Tardieu wrote:
> | > The removal of this code doesn't change the outcome of GCC GNAT and
> | > ACATS tests. Can someone explain if this code is useful?
> | 
> | Note that GCC ACATS contains only executable tests, this
> | kind of patch might affect some ACATS B non executable tests.
> 
> Right.
> 
> Do you know if there are any plan to incorporate the ACATS 3.0 test
> suite into GCC? This looks like quite a large job (I've never played
> with ACATS except running them from within GCC test suite) to integrate
> them.

Integrating new executable tests from ACATS new releases is a small and
easy job: just commit the files in the correct chapter directory. The
hardest part is figuring out what is the current official list of
tests :).

When I ported ACATS for GCC I chose not to include the non executable
B tests because are checking for error messages and will need
maintenance each time error messages or error recovery code path change
in the compiler.

May be such required changes are less frequent now and the "split" base
at AdaCore is more stable: a B test source code can be split by the
compiler vendor if the compiler is loosing track of errors in one file
after some point, most ACATS B tests check for dozens of errors at once.

May be Robert has some up to date statistics?

Laurent