David,
It seems the remaining linkage problem with llvm36 is due to the
reduced linkage on -lcurses in 3.6. The prior llvm releases seems to
link all of the shared libraries with -lcurses but this has been
dramatically reduced in 3.6 to just libLLVMSupport.dylib in
stage1/lib. The linkage of LLVMPolly.so fails because of the missing
linkage on -L%p/lib -lcurses causes the system curses to be indirectly
dragged in by the linker…
ld: in '/usr/lib/libncurses.5.dylib', file was built for unsupported
file format ( 0xce 0xfa 0xed 0xfe 0x 7 0x 0 0x 0 0x 0 0x 3 0x 0 0x 0
0x 0 0x 6 0x 0 0x 0 0x 0 ) which is not the architecture being linked
(x86_64): /usr/lib/libncurses.5.dylib for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
I haven't had much luck yet identifying exactly what in the 3.6
CMakefiles causes bLLVMSupport.dylib to explicitly link against
-lcurses but that change needs to be applied to the LLVMPolly linkage
in polly-3.6.0.src/lib/CMakeLists.txt.
Jack
On Mon, Apr 6, 2015 at 10:00 PM, Jack Howarth <howarth.at.f...@gmail.com> wrote:
> David,
> I think I see the defect in the llvm3x packaging. You need to
> have something like…
>
> Index: llvm35.info
> ===================================================================
> RCS file: /cvsroot/fink/experimental/jwhowarth/main/finkinfo/llvm35.info,v
> retrieving revision 1.26
> diff -r1.26 llvm35.info
> 392c392
> < STAGE1_LD_FLAGS="${STAGE1_LD_FLAGS[@]}"
> ---
>> STAGE1_LD_FLAGS="${STAGE1_LD_FLAGS[@]} ${LDFLAGS[@]}"
>
> The current code is effectively clearing the LDFLAGS which explains
> that weird linkage on the system libncurses in stage1.
> Jack
>
> On Mon, Apr 6, 2015 at 9:36 PM, Jack Howarth <howarth.at.f...@gmail.com>
> wrote:
>> David,
>> Have you attempted to rebuild the current llvm35-3.5.1-1
>> packaging against the current cmake 3.2.1 in fink on 10.7? The newer
>> cmake 3.2.1 appears to be the trigger for the build failures on 10.7.
>> Reverting back to cmake 3.1.2 eliminates the build failures.
>> One aspect of the llvm3x package build really confuses me. The
>> successful stage1 build seems to link entirely against the system
>> libncurses whereas the failing stage1 build picks up -L/sw/lib in the
>> libLLVMSupport.dylib linkage. Strangely both stage2 and stage3 (in the
>> successful builds) are linking entirely against fink's libncurses.
>> However I don't quite see what toggles this behavior. At first glance,
>> all three stages should be linking against the fink libraries for
>> libncurses, no?
>> Jack
------------------------------------------------------------------------------
BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
Develop your own process in accordance with the BPMN 2 standard
Learn Process modeling best practices with Bonita BPM through live exercises
http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_
source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF
_______________________________________________
Fink-devel mailing list
Fink-devel@lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel
