On Apr 12, 2007, at 1:54 PM, Nicholas Clark wrote:
On Thu, Apr 12, 2007 at 01:50:09PM -0500, Joshua Isom wrote:
On Apr 12, 2007, at 9:29 AM, Nicholas Clark wrote:
My view of this is something along these lines. You can use any
function you want at all, but if it's not documented as part of the
supported API, there's nothing saying I won't purposely break your
code
in the next release. In big bold red letters of course.
Particularly given that on some platforms (such as Win32) API
functions need
to be exported explicitly, resulting in modules written on *nix
platforms
not working on Win32.
The problem is that between now and that next release, someone
inevitably
writes something on Linux that doesn't work on Win32, and they blame
everyone
but themselves.
I'd much prefer to nip the problem in the bud by preventing the code
from
ever being shipped.
Nicholas Clark
I think it's been a while since we had many failure free smokes. Is
there/does someone know of a better "smoking" module to make it easily
to test failures and if it's architecture specific, or operating system
specific, or both. There's a few problems with freebsd on amd64 which
may largely be configuration(freebsd is amd64, not x86_64).
Portability on parrot is more than a problem of exported symbols at the
moment. I think we've become overly accustomed to having test
failures, but some installer scripts won't install if there's a single
failure, which may be something trivial.