On Oct 18, 2011, at 16:53, Andrea D'Amore wrote:
>> As I mentioned in the ticket, I'm not comfortable with a universal variant
>> that hardcodes the architectures, nor one that's specific to an OS version.
>> Why does this universal variant have to be specific to Lion? Why can't it
>> respect the ${universal_archs} variable?
>
> As I replied I'm not able to test on 10.6, I can't build ppc and I
> know how easy it is to break things when changing system version.
> In the same ticket I asked you to remove the reinplace for ppc and
> test the variant port if you could. On darwin 11 ppc target arch has
> to be removed for sure.
On Darwin 11 (and on Darwin 10, really) it wouldn't make sense for the user to
list ppc in universal_archs in macports.conf; it would be a user error for them
to do so.
I don't really have time to test this on ppc. I was merely pointing out code
which did not conform to our usual portfile writing methods. Most portfile
authors do not have every OS X version and architecture available to test on;
they simply assume it will work on other platforms, which it often will. We
shouldn't fill portfiles with conditions restricting things to only the tested
platforms. Rather, we should only introduce such conditionals after it has been
proven that something does not work on a given OS, or if the program's
documentation says it will not work on a given OS.
In this case, the configure script hardcodes the universal arch flags to use.
So the fix should be to replace that hardcoded list with the value returned
from the [get_canonical_archflags] procedure. That should theoretically work on
any OS version with any combination of universal architectures.
_______________________________________________
macports-dev mailing list
[email protected]
http://lists.macosforge.org/mailman/listinfo.cgi/macports-dev
