p7zip @15.09 won't compile on PPC Leopard, Mac OS X 10.5.8, because of missing /usr/bin/llvm-gcc

[ballapete (Peter "Pete" Dyballa)]

:debug:build Assembled command: 'cd "/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09" && /usr/bin/make -w all3'
:debug:build Executing command line:  cd "/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09" && /usr/bin/make -w all3 
:info:build make: Entering directory `/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09'
:info:build mkdir -p bin
:info:build /usr/bin/make -C CPP/7zip/Bundles/Alone all
:info:build make[1]: Entering directory `/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09/CPP/7zip/Bundles/Alone'
:info:build /usr/bin/llvm-gcc -c -I. -I../../../../C -I../../../myWindows -I../../../include_windows -I../../.. -m64 -O -DENV_MACOSX -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_REENTRANT -DENV_UNIX -arch ppc  -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_REENTRANT -DENV_UNIX -DBREAK_HANDLER -DUNICODE -D_UNICODE -DUNIX_USE_WIN_FILE   ../../../../C/7zCrc.c
:info:build make[1]: /usr/bin/llvm-gcc: Command not found
:info:build make[1]: *** [7zCrc.o] Error 127
:info:build make[1]: Leaving directory `/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09/CPP/7zip/Bundles/Alone'
:info:build make: *** [7za] Error 2
:info:build make: Leaving directory `/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09'
:info:build Command failed:  cd "/opt/local/var/macports/build/_opt_mports_trunk_dports_archivers_p7zip/p7zip/work/p7zip_15.09" && /usr/bin/make -w all3 
:info:build Exit code: 2
:error:build Failed to build p7zip: command execution failed
:debug:build Error code: CHILDSTATUS 49214 2

I have:

lrwxr-xr-x 1 root wheel      7  2. Dez 2013  /usr/bin/gcc -> gcc-4.2
-rwxr-xr-x 1 root wheel  93088  5. Feb 2009  /usr/bin/gcc-4.0
-rwxr-xr-x 1 root wheel 105680  7. Jul 2009  /usr/bin/gcc-4.2
-rwxr-xr-x 1 root wheel 369408  5. Feb 2009  /usr/bin/i686-apple-darwin9-gcc-4.0.1
-rwxr-xr-x 1 root wheel 505120  7. Jul 2009  /usr/bin/i686-apple-darwin9-gcc-4.2.1
-rwxr-xr-x 1 root wheel 369408  5. Feb 2009  /usr/bin/powerpc-apple-darwin9-gcc-4.0.1
-rwxr-xr-x 1 root wheel 513344  7. Jul 2009  /usr/bin/powerpc-apple-darwin9-gcc-4.2.1

-rwxr-xr-x 1 root admin     32 17. Jan 2013  /Developer/usr/bin/llvm-gcc -> ../llvm-gcc-4.2/bin/llvm-gcc-4.2
lrwxr-xr-x 1 root admin     32 17. Jan 2013  /Developer/usr/bin/llvm-gcc-4.2 -> ../llvm-gcc-4.2/bin/llvm-gcc-4.2
-rwxrwxr-x 1 root admin 109808 24. Mär 2009  /Developer/usr/llvm-gcc-4.2/bin/llvm-gcc-4.2

[ryandesign (Ryan Carsten Schmidt)]

p7zip is evidently using a hard coded compiler. It should be changed to be UsingTheRightCompiler.

[ballapete (Peter "Pete" Dyballa)]

main.log

[Schamschula (Marius Schamschula)]

Cc Me!

[ballapete (Peter "Pete" Dyballa)]

I created a sym-link /usr/bin/llvm-gcc -> /Developer/usr/bin/llvm-gcc-4.2 (ans also /usr/bin/llvm-gcc -> /Developer/usr/bin/llvm-gcc-4.2) which allowed me to have it built. /usr/bin now has these llvm, gcc, and g++ related sym-links and files:

lrwxr-xr-x  1 root   wheel            7  2 Dez  2013 cc -> gcc-4.2
lrwxr-xr-x  1 root   wheel            7  2 Dez  2013 g++ -> g++-4.2
-rwxr-xr-x  1 root   wheel        93088  5 Feb  2009 g++-4.0
-rwxr-xr-x  1 root   wheel       105680  7 Jul  2009 g++-4.2
lrwxr-xr-x  1 root   wheel            7  2 Dez  2013 gcc -> gcc-4.2
-rwxr-xr-x  1 root   wheel        93088  5 Feb  2009 gcc-4.0
-rwxr-xr-x  1 root   wheel       105680  7 Jul  2009 gcc-4.2
-rwxr-xr-x  1 root   wheel       373504  5 Feb  2009 i686-apple-darwin9-g++-4.0.1
-rwxr-xr-x  1 root   wheel       513344  7 Jul  2009 i686-apple-darwin9-g++-4.2.1
-rwxr-xr-x  1 root   wheel       369408  5 Feb  2009 i686-apple-darwin9-gcc-4.0.1
-rwxr-xr-x  1 root   wheel       505120  7 Jul  2009 i686-apple-darwin9-gcc-4.2.1
lrwxr-xr-x  1 root   wheel           31  2 Jan 14:39 llvm-g++ -> /Developer/usr/bin/llvm-g++-4.2
lrwxr-xr-x  1 root   wheel           31  2 Jan 14:39 llvm-gcc -> /Developer/usr/bin/llvm-gcc-4.2
-rwxr-xr-x  1 root   wheel       377616  5 Feb  2009 powerpc-apple-darwin9-g++-4.0.1
-rwxr-xr-x  1 root   wheel       513344  7 Jul  2009 powerpc-apple-darwin9-g++-4.2.1
-rwxr-xr-x  1 root   wheel       369408  5 Feb  2009 powerpc-apple-darwin9-gcc-4.0.1
-rwxr-xr-x  1 root   wheel       513344  7 Jul  2009 powerpc-apple-darwin9-gcc-4.2.1

Is port select --set gcc also manipulating these files, particularly when it sets apple-gcc42?

[khepler]

When ticket #49788 was applied, the Portfile switched from using makefile.macosx_32bits to makefile.macosx_llvm_64bits. That defeated the compiler override, which means right now no architecture is using the "right" compiler.

Attaching a cross-platform patch.

[khepler]

Cc Me!

[khepler]

use the right compiler (corrected)

[kurthindenburg (Kurt Hindenburg)]

Cc Me!

[kurthindenburg (Kurt Hindenburg)]

Thanks - your patch didn't apply cleanly but I think I got it correct. reopen if there are still issues.

r145156

[khepler]

Builds on Leopard/PPC.

What was wrong with the patch?

[kurthindenburg (Kurt Hindenburg)]

It didn't apply cleanly - not sure if you were using an old Portfile or what.

[khepler]

Ah, looks like the Portfile was updated half an hour after I submitted the patch.

[jmroot (Joshua Root)]

This change adds unnecessary complexity and introduces some regressions.

• The makefile is being selected based on build_arch, which is incorrect in the case of a universal build. It was correct before because the -m64 was stripped out and our own archflags added.
• makefile.macosx_gcc_64bits has no significant differences from makefile.macosx_llvm_64bits.
• The only significant difference in makefile.macosx_gcc_32bits_ppc is that it has a hardcoded -mmacosx-version-min=10.4, which is wrong.
• makefile.macosx_gcc_32bits_asm also has the hardcoded -mmacosx-version-min=10.4.
• It also adds an undeclared dependency on nasm.
• The one possibly useful bit is enabling asm on i386. But this should only be done when i386 is the only arch being built for.

[khepler]

You are correct; the Portfile half an hour newer than my patch probably would work fine and is certainly simpler.

What does build_arch look like in universal build?

[jmroot (Joshua Root)]

Replying to khepler@…:

You are correct; the Portfile half an hour newer than my patch probably would work fine and is certainly simpler.

Reverted in r145182. If anyone wants to enable asm for i386, please open a new ticket for that.

What does build_arch look like in universal build?

It's always the value that is set in macports.conf. It's not used in universal builds though, universal_archs is.

Anyway, this ticket is resolved now, so we should take further discussion elsewhere.