On Feb 12, 2006, at 07:41 , Ralf Wildenhues wrote:
* D. Walsh wrote on Sun, Feb 12, 2006 at 01:13:39PM CET:
On Feb 12, 2006, at 05:17 , Ralf Wildenhues wrote:
[ moving from the libtool list ]
As a test, I tried this patch on Mac OSX, sorry to say it wasn't a
happy camper.
Could you be more specific? For example, posting the resulting errors
would be great.
Missing DLL related issues (I don't have any dll files on my Mac and
it's not a PPC Mac) but this could be because it's currently not a
recognized vendor/cpu combination and I have to massage the configure
flags to compile many GNU packages.
I have to agree, some of the proposed manipulations are done too late
in the game and should be done in the deplib search.
One concern (with Apple switching to Intel processors) is dual
architecture binaries and cross-compilations problems are going to
occur if proper care isn't taken to take into account which CPU is
used/available and if the environment is set to generate universal
binaries.
Erm, what does this have to do with the proposed MSVC patch at all?
Currently the patch has negative effects on an Intel Mac but will
likely be corrected with updating.
I can easily see that Libtool will need to be adjusted to take this
into account for OSX/x86 support, but I don't really see the
connection
to w32 issues. They are pretty orthogonal.
This is assumed, currently on an Intel Mac, part of it is assuming
w32 while I've specifically set my environment to generate universal
binaries.
More than likely, adjustments to consider Mac->X86 will resolve many
issues because I've done a little further testing myself.
I looked at applying the Apple Mac->Intel patch and it can be applied
to the latest version (with a little manual help) and this seems to
make the MSVC patch appears to be non-existent which is the way it
should be.
Cheers,
Ralf
