On Thu, 13 Sep 2007, James E Keenan wrote:
> Andy Dougherty wrote:
>
> > (If cc_build and cc_run do get meaningful exit codes, inter::progs needs to
> > be revisited anyway to actually use those codes.)
>
> If you do revisit config/inter/progs.pm, please bear in mind that I have
> written some unit tests for that package:
>
> [parrot] 500 $ ls t/configure/*progs* | cat
> t/configure/107-inter_progs.01.t
> t/configure/107-inter_progs.02.t
> t/configure/107-inter_progs.03.t
> t/configure/107-inter_progs.04.t
To be blunt, I am more interested at the moment in whether it's trying
to do something sensible than whether or not its actual current behavior
is being tested. Testing the existing behavior assumes that the existing
behavior is what the program should actually be doing.
Trivial example: Configure.pl currently supports many incompatible ways
to say "no" (excerpts from Configure.pl --help)
--nomanicheck Don't check the MANIFEST
--cgoto=0 Don't build cgoto core - recommended when short of mem
--without-gdbm Build parrot without GDBM support
--icu-config=none Can be used to disable the autodetection feature.
It will also be disabled if you specify any other
of the following ICU options.
Rather than testing that each one is handled correctly, it would make
more sense to design and implement a single way to say "no". Then you
can test the daylights out of that.
Sorry if this sounds too grumpy. Having a testing procedure, framework,
and culture is indeed useful, and your efforts there are appreciated.
But having a good design would also be useful.
--
Andy Dougherty [EMAIL PROTECTED]
