Justin Erenkrantz wrote:
On Tue, Jan 10, 2006 at 10:11:04AM -0700, Martin Sebor wrote:
[...]
% nm /usr/lib/gcc/powerpc-apple-darwin8/4.0.0/libstdc++.dylib | egrep
"__gxx_personality_v0|__cxa_throw|__cxa_allocate" | grep " T "
9550da34 T ___cxa_allocate_exception
9550ec80 T ___cxa_throw
9550e5d0 T ___gxx_personality_v0
So, does that mean Apple did something they shouldn't have done?
These symbols (basically the implementation of the C++ runtime ABI)
are actually expected to be defined in both libstdc++ and libsupc++
so this in itself is not the smoking gun. The absence of libsupc++
does make me suspect that they might have "simplified" things by
not providing libsupc++. We'll need to confirm this with Apple.
Let me ask Howard Hinnant, their new libstdc++ guy.
I will note that I have:
gcc version 4.0.0 (Apple Computer, Inc. build 5026)
The current version appears to be:
gcc version 4.0.1 (Apple Computer, Inc. build 5247)
4.0.1 produces the same error and has the same symbols in libstc++, FWIW.
It's Andrew's old laptop :( We need to get a more recent OS X system
for testing. Do you happen to know if Apple has some kind of a "test
drive" program similar to HP's (http://www.testdrive.hp.com/)?
The ASF has access to a few Mac OS X machines running 10.4. Send me your
ssh public key off-list and I can get you access. -- justin
Okay, will do. Thanks!
Martin
