Peter, Gary and other libtoolers,
I wasn't sure George got back to you on this.
With some more testing, apparently MACOSX_DEPLOYMENT_TARGET set to 10.3
*does* work for us. I suspect something didn't get rebuilt the first
time around.
I think we all agree that there is a problem in echoing '${wl}
Hmm, as far as I can tell, it is being passed by GNU libtool to g++
successfully, it is just echo'ed incorrectly. But I believe you and
will try and reproduce it this weekend.
That said, do you mean running libtool with MACOSX_DEPLOYMENT_TARGET
set to 10.3,
or is it sufficient to just have that
George Feinberg wrote:
Please note that if I successfully pass "-flat_namespace -undefined
suppress" to the
linker, things work just fine. This used to be the case, with DB 4.2.52,
which used an older libtool. There is still a bug in libtool in that these
are not properly passed on OS X.
Hmm, as
George Feinberg wrote:
Note that ALL 3 of these libraries load all of the core DB .o files,
and therefore
duplicate those symbols.
This is the problem. Tcl, for some reason, uses NSAddImage to load
objects at runtime on darwin/Mac OS X. NSAddImage, and NSAddLibrary,
are unable to load objects pr
George Feinberg wrote:
Note that ALL 3 of these libraries load all of the core DB .o files, and
therefore
duplicate those symbols.
This is the problem. Tcl, for some reason, uses NSAddImage to load objects
at runtime on darwin/Mac OS X. NSAddImage, and NSAddLibrary, are unable to
load objects pr
On Oct 13, 2004, at 3:35 AM, Gary V. Vaughan wrote:
Don, George: On what release of OS X did you find this bug? What
value of MACOSX_DEPLOYMENT_TARGET (if any) was set in the environment?
I'm running 10.3 on a PowerBook G4.
If I have no MACOSX_DEPLOYMENT_TARGET set, and run configure without
any
Hi Peter,
Peter O'Gorman wrote:
Gary V. Vaughan wrote:
Isn't the problem report -Wl,suppress vs suppress, rather than the wonky
${wl} in the $show?
I don't know. I awoke this morning fully convinced it was Friday (turned
out to be Wednesday), so I may have misread :-).
Odd that, Val (my partner)
Gary V. Vaughan wrote:
Isn't the problem report -Wl,suppress vs suppress, rather than the wonky
${wl} in the $show?
I don't know. I awoke this morning fully convinced it was Friday (turned out
to be Wednesday), so I may have misread :-). However, it worked for me just
now when I tried it from the
Hi Peter,
Thanks for the fast response.
Peter O'Gorman wrote:
A secondary issue is that changing this to
"-Wl,-flat_namespace -Wl,-undefined -Wl,suppress" does not work,
and the string passed to g++ must be
"-Wl,-flat_namespace -Wl,-undefined suppress"
While I have seen this issue before, I have n
Hi,
[sleepycat support removed, libtool added]
A secondary issue is that changing this to
"-Wl,-flat_namespace -Wl,-undefined -Wl,suppress" does not work,
and the string passed to g++ must be
"-Wl,-flat_namespace -Wl,-undefined suppress"
While I have seen this issue before, I have never seen a fai
10 matches
Mail list logo
