[SDL] [Fwd: Re: [fltk.development] Re: [Library] r5178-inbranches/branch-1.1:. src]

[E. Wing]

>> It is not possible to do this - as I've mentioned in several other
>> posts in this thread, only the Intel installs of MacOS X include
>> Universal libraries and frameworks, so you have to build on Intel
>> and then test/validate on PPC.
>>
>> This will change in MacOS X 10.5...
>
>
> Is this correct ?

No, this is not correct. (There would have been riots by Mac
developers if this was the case.) Xcode 2.1+ ships with the tool set
you need to build Universal Binaries. They basically modified gcc to
access its cross compiler features and leveraged the dormant fat
binaries features left over from the NeXT days. Xocde includes the
10.4 Universal SDK which basically contains copies of all the standard
OS X frameworks, but built as Universal binaries so you can build
against them to produce your own Universal binaries. If you don't
believe me, check out
/Developer/SDKs/MacOSX10.4u.sdk/System/Library/Frameworks and you'll
see the same list of frameworks as in your /System/Library/Frameworks.
But if you examine the binaries in the frameworks (using Get Info or
lipo), you'll see they are all built as Universal.

Testing is another matter though. You can't really "test" if the i386
code works correctly from using only a PPC machine. If your code is at
a high level where none of your code is architecture/endian dependent,
you can try "trusting" that the underlying frameworks got stuff right
, in which case testing on just PPC might be sufficient for you if you
have a huge amount of faith in everything. If you have an Intel
machine, you have the possibility of running through Rosetta to test
PPC, but you have to have faith that Rosetta behaves exactly like
running on a real PPC.

-Eric