Plans for Lion

Sticking with already-public info...there are some design decisions and migration issues we need ASAP to
enable further testing and preparations for 10.7. Here are my notes:

1. Architecture. Fink on 10.7 will be x86_64 only (not supporting i386). That will solve a ton of
perl-module (and other) packaging issues involving forcing use of the "non-default" arch
consistently. 

2. "Upgrading" from 10.6 will not work cleanly because apple apparently does not have
/usr/bin/perl5.10.0 at all (hardcoded #! interp in fink core). Therefore users need to reinstall from
scratch (and can use 'dpkg' but not 'fink' to dump a list of installed packages before doing so). Users with
arch=i386 would not be able to upgrade anyway even with altering #! since we don't support that arch on
10.7. 

3. The system-perl interp may not even have a fully versioned name. That makes it unusable (and potentially
not even unchanging!) for perlX.X.X varianting. 

4. Given #2 and #3, I think we should have our own perl interp for fink core instead of using system-perl. We
did this on a previous distro. It gives fuel to "fink is duplicating so much stuff I already have!" critics,
but it also solves arch-consistency and pathname sanity (the "yes, but it's the solution that works"
response). 

5. Problem: We have tons of legacy support for older OS X in packages (variants that don't exist on newer
distros, hacks for distro-specific bugs). We also have some packages blocked from upgrading because
they cannot be used on older distro and it's painful to fork a low-level library among distros (means the
whole dep tree from that point has to fork in order to allow versioned dependencies). 

6. We have to decide which perl- and python- variants we want to bring forward. Historically we've kept >=
"whatever came with previous OS X system". That was especially important for perl, so that things that
used system-perl could continue to work seemlessly.