upgrade observations

Neil kngspook at gmail.com
Sun Jul 5 03:54:29 PDT 2009


On Sat, Jul 4, 2009 at 2:49 PM, Rich Morin <rdm at cfcl.com> wrote:

> At 17:01 -0400 7/4/09, Brandon Allbery wrote:
> > On Jul 4, 2009, at 11:15 , Rainer Müller wrote:
> > I read that as "relocatable" in the object/symbol table sense
> > (which I'm not sure applies to PPC).
>
> Strictly speaking, it may not.  What I was thinking about, in
> any case, was the issue of linking compiler output files (eg,
> foo.o) with libraries (eg, bar.a) to produce executable binary
> images (eg, foo).
>
> I don't really care if MacPorts supports cross-compilation, let
> alone produces universal binaries.  The key issue is whether the
> user gets a WTF experience after moving (and even using, for a
> while) the /opt tree.
>
> A really simple solution would be for the port command to look
> at the current architecture type, compare it to what it expects,
> and issue a simple nastygram if it differs.  Like:
>
>  Cannot mix ppc and x86 architectures - bailing out...
>

Well, having it give an error message would be a good idea if someone tries
something like what you tried.

But I see little reason to support being able to move the tree from one
machine to another.

At worst, move the $PREFIX/etc directory, where all the prefs are (/should
be).

-N.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macosforge.org/pipermail/macports-users/attachments/20090705/50967b05/attachment.html>


More information about the macports-users mailing list