On Sat, 17 Apr 2010, Marc Espie wrote:
> there's been some discussion about it. The idea is to move some information
> from dynamic handling to static visibility.
>
> Namely, a pattern such as
> .if ${ARCH} == "sparc64"
> BROKEN = "doesn't work on sparc64, we don't know why"
> .endif
> has low visibility. Anyone who works on the ports tree outside of sparc64
> won't see it.
>
> It's also totally different in intent from ONLY_FOR_ARCHS and NOT_FOR_ARCHS:
> something that is completely designed to not work on some architecture is
> not really "broken" as "needing lots and lots of work".
>
> For instance, opera is a binary port, saying BROKEN on !i386 is a nonsense.
>
> "Portable" languages such as java which don't work at all outside of a very
> limited set of architectures are not broken either (well, one could argue
> that they're completely broken in fact...)
>
> The neat thing with BROKEN-* is that, now that I've finished the required
> changes, it will show up in sqlports (I had not thought about how to do
> it beforehand, so it took a few weeks to explain arch-dependent variables
> to dump-vars and to mksqlitedb).
>
>
> So, now, any enterprising soul can just open the sqlports database with
> their favorite sqlitebrowser, peer through the BROKEN table, and have a try
> at fixing the issues.
So you basically fucked our last excuse for holidays...
THANK YOU MARC, THANK YOU!
--
Antoine
