The below exchange on p5p made me wonder if there's a need for an API
that issues warnings or informational messages in addition to the
current pass/fail alternatives. Kind of like a "shouldbe" alternative
to "is".
This would allow a probe for possible gotchas to be made in a release
before actully implementing something based on those gotchas, assuming
people could be convinced to mail in feedback from a message that says:
ok 12 # Whoops! Got 'foo', should be 'bar' at t/whumpus.t, line 23
Good/Bad/Useless?
- Barrie
On Thu, Jul 26, 2001 at 08:54:40AM +0100, Nick Ing-Simmons wrote:
> Kurt D . Starsinic <[EMAIL PROTECTED]> writes:
> >Hello,
> >
> > The following patch:
> >
> > Adds a test, h2ph_thorough, that compares constants available in
> > POSIX.pm and Socket.pm to their equivalents as generated by h2ph.
> >
> > Reveals many bugs in h2ph.
> ...
> >
> >
> > I expect the Average Platform to get three failures on the test.
> >These are old bugs; we just didn't have tests for them before. I'm
> >working on fixing them.
>
> Not applied (yet) - I am a little wary of adding fails - what is guestimate
> as when the fixes for an "Average Platform" will be ready?
> Also what about non-average platforms where POSIX and Socket or
> "equivalent" .h files may be "odd" ?
>
> --
> Nick Ing-Simmons
> http://www.ni-s.u-net.com/
>
