On Tue, Jun 07, 2005 at 09:01:02AM +0900, Peter O'Gorman wrote:
> 
> I advise against the fat approach, I know Rob Braun tried it at some point
> and had issues (I cc'ed him).

My experience has mostly been with ppc and i386 fat building.
Going all fat isn't terrible.  It'll cause problems in a few
packages (notably gmp), but for the most part its not that bad.
However, it does raise the bar a bit for port maintainers.
The real headache is mixing fat and thin.  It should be all or
nothing.  Having dependencies on specific architectures (required
for building and such).  Building fat is fundamentally different
than building thin for each architecture and merging the results.
Neither is "better", just different.  In some cases building fat
may be the desired way, other times it may be better to build thin
and intelligently merge the results.  Some of the generated files 
may be different.  It's not just merging fat binaries.

Anyway, I'd suggest that if supporting fat is a goal, going 
all fat or no fat would be my vote.  Going all fat will be a
hassle but doable.  Adding/removing an architecture in the future
would still raise the partial fat problems.

In any case, fat is "hard", and if it is a goal, we can start
keeping a list of known issues somewhere, deciding what will
and won't be solved immediately, and then start prototyping it
to find other issues.

Rob


-------------------------------------------------------
This SF.Net email is sponsored by: NEC IT Guy Games.  How far can you shotput
a projector? How fast can you ride your desk chair down the office luge track?
If you want to score the big prize, get to know the little guy.  
Play to win an NEC 61" plasma display: http://www.necitguy.com/?r=20
_______________________________________________
Fink-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/fink-devel

Reply via email to