On Wed, Mar 02, 2011 at 02:27:12PM +0100, Pascal Stumpf wrote: > Perl 6 on OpenBSD needs some love, so I ported the most recent stable > version of the parrot, plus the Rakudo Perl6 compiler. The parrot passes > all currently implemented regression tests on amd64, and rakudo passes > all but five subtests in an extensive spectest run. > > The parrot Makefile indicates build failures on macppc and sparc64, so > it would be nice to know if those platforms (or, for that matter, > anything else than i386 and amd64) work or not.
Fwiw, it builds and regresses fine on ppc, and rakudo is still building there. On sparc64, parrot builds but lots of regress tests fails, and rakudo fails to build: /usr/local/bin/parrot-nqp --target=pir --output=src/gen/perl6-grammar.pir --encoding=utf8 src/Perl6/Grammar.pm Segmentation fault (core dumped) See regress logs at: http://rhaalovely.net/~landry/stuff/parrot-3.0.0-macppc-regress http://rhaalovely.net/~landry/stuff/parrot-3.0.0-sparc64-regress If you can forward the results upstream, or ask them if they plan to support sparc64 someday.... Now, portswise.. > -SHARED_LIBS= parrot 0.0 > +SHARED_LIBS= parrot 0.1 You checked the api, right ? I doubt there was only function additions since 1.2.0.. The .if chunk to switch a configure arg only on ppc should go away, as it's now and unknown option. > +src/ > +src/parrot/ > +src/parrot/3.0.0/ > +src/parrot/3.0.0/pmc/ > +src/parrot/3.0.0/pmc/addrregistry.dump Err.. no way, this is outside FHS. If the files are needed by parrot, patch it so that it finds them in a correct location (dunno, share/parrot/src maybe), if not them remove them. As for the rakudo port itself, you should relax LIB_DEPENDS, lang/parrot=3.0.0 is too strict, imo >=3.0.0 would be better. Landry
