boost @1.59.0: won't install on 10.5.8 Leopard

[stuchalk (Stuart Chalk)]

In trying to upg source-lighlight, boost won't install on leopard. Error file attached...

[skoro.ru@…]

OS X 10.5.8 / boost @1.59.0

Same issue seen here:

:info:build ...failed darwin.compile.asm bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/asm/make_ppc32_sysv_macho_gas.o...

Previous versions installed and working:

boost @1.57.0_1+no_single+no_static+python27
boost @1.58.0_1+no_single+no_static+python27
boost @1.58.0_2+no_single+no_static+python27
boost @1.58.0_3+no_single+no_static+python27 (active)

[ryandesign (Ryan Carsten Schmidt)]

This does not sound like a MacPorts-specific problem. Could one of you please ​report it to the developers of boost?

[ballapete (Peter "Pete" Dyballa)]

On my PPC system with Mac OS X 10.5.8 I get this info on the failure:

Command failed:  cd "/opt/local/var/macports/build/_opt_mports_trunk_dports_devel_boost/boost/work/boost_1_59_0" && /opt/local/var/macports/build/_opt_mports_trunk_dports_devel_boost/boost/work/boost_1_59_0/b2 -d2 --layout=tagged --debug-configuration --user-config=user-config.jam -sBZIP2_INCLUDE=/opt/local/include -sBZIP2_LIBPATH=/opt/local/lib -sEXPAT_INCLUDE=/opt/local/include -sEXPAT_LIBPATH=/opt/local/lib -sZLIB_INCLUDE=/opt/local/include -sZLIB_LIBPATH=/opt/local/lib -sICU_PATH=/opt/local variant=release -j1 architecture=power address-model=32 --disable-long-double link=shared threading=multi 
Exit code: 1
Error: Failed to build boost: command execution failed
DEBUG: Error code: CHILDSTATUS 15744 1
DEBUG: Backtrace: command execution failed
    while executing
"system {*}$notty {*}$nice $fullcmdstring"
    invoked from within
"command_exec build"
    (procedure "portbuild::build_main" line 8)
    invoked from within
"$procedure $targetname"
Error: See /opt/local/var/macports/logs/_opt_mports_trunk_dports_devel_boost/boost/main.log for details.
DEBUG: invalid command name "::ui_init"
    while executing
"::ui_init $priority $prefix $channels($priority) {*}$args"
    ("uplevel" body line 2)
    invoked from within
"uplevel 1 $body"

[ballapete (Peter "Pete" Dyballa)]

Cc Me!

[ballapete (Peter "Pete" Dyballa)]

Pete's main.log

[ballapete (Peter "Pete" Dyballa)]

Replying to ryandesign@…:

This does not sound like a MacPorts-specific problem. Could one of you please ​report it to the developers of boost?

I could reproduce your failure by having port upgrade some other ports of which I thought that they do nor depend on boost. So I did not perform port clean boost and therefore the process started to continue where it failed before. The report

Error: Failed to build boost: command execution failed
DEBUG: Error code: CHILDSTATUS 1632 1
DEBUG: Backtrace: command execution failed
    while executing
"system {*}$notty {*}$nice $fullcmdstring"
    invoked from within
"command_exec build"
    (procedure "portbuild::build_main" line 8)
    invoked from within
"$procedure $targetname"
Error: See /opt/local/var/macports/logs/_opt_mports_trunk_dports_devel_boost/boost/main.log for details.
DEBUG: can not find channel named "debuglog"
    while executing
"tcl::puts debuglog {DEBUG: Starting logging for boost}"
    invoked from within
"catch "tcl::puts $args""

certainly is from port.

[ryandesign (Ryan Carsten Schmidt)]

Stuart, thank you for ​reporting the bug to the developers of boost, but the log you attached there (and here) was not from a ​clean build attempt; the developers may want to see a complete log file. You should also give them a link to this bug report so they know it has already been reported to MacPorts.

[ryandesign (Ryan Carsten Schmidt)]

Peter, there is a bug in MacPorts (I don't know if we already have a ticket for it) that often causes it to print the wrong error message at the end of a failed build. That may be what you're showing. The real reason why boost failed to build would be earlier in the log and needs to be solved by the boost developers.

[ballapete (Peter "Pete" Dyballa)]

Replying to ryandesign@…:

I reported the bug to the boost developers: ​https://svn.boost.org/trac/boost/ticket/11578.

[ryandesign (Ryan Carsten Schmidt)]

Thanks, but Stuart already reported it.

Has duplicate #48694.

[dgonyier (Dwaine Gonyier)]

Cc Me!

[ryandesign (Ryan Carsten Schmidt)]

Has duplicate #49014. There has been no response so far to the upstream bug report.

[ryandesign (Ryan Carsten Schmidt)]

A solution was proposed here: ​https://svn.boost.org/trac/boost/ticket/11603. It says that

     stw  r5, 236(%r3)

is a typo and should instead be

     stw  r5, 236(r3)

I found ​this reference that says that the "%" is used in the GNU assembler, but not the Apple assembler which is what we are using.

I have not yet had a chance to test this proposed solution.

[ballapete (Peter "Pete" Dyballa)]

Replying to ryandesign@…:

A solution was proposed here: ​https://svn.boost.org/trac/boost/ticket/11603. It says that

     stw  r5, 236(%r3)

is a typo and should instead be

     stw  r5, 236(r3)

This is a good start! It leads to this:

darwin.compile.asm bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/asm/make_ppc32_sysv_macho_gas.o

    "/usr/bin/g++-4.2" -x assembler-with-cpp -O3 -finline-functions -Wno-inline -Wall -dynamic -gdwarf-2 -fexceptions -fPIC -arch ppc -DBOOST_ALL_NO_LIB=1 -DBOOST_CHRONO_DYN_LINK=1 -DBOOST_CONTEXT_DYN_LINK=1 -DBOOST_CONTEXT_SOURCE -DBOOST_DISABLE_ASSERTS -DBOOST_SYSTEM_DYN_LINK=1 -DBOOST_SYSTEM_NO_DEPRECATED -DBOOST_THREAD_BUILD_DLL=1 -DBOOST_THREAD_POSIX -DBOOST_THREAD_USE_DLL=1 -DNDEBUG -I"." -c -o "bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/asm/make_ppc32_sysv_macho_gas.o" "libs/context/src/asm/make_ppc32_sysv_macho_gas.S"

darwin.link.dll stage/lib/libboost_context-mt.dylib

    "/usr/bin/g++-4.2" -dynamiclib -Wl,-single_module -install_name "/opt/local/lib/libboost_context-mt.dylib"  -o "stage/lib/libboost_context-mt.dylib" "bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/asm/make_ppc32_sysv_macho_gas.o" "bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/asm/jump_ppc32_sysv_macho_gas.o" "bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/posix/stack_traits.o" "bin.v2/libs/context/build/darwin-4.2.1/release/threading-multi/execution_context.o" "bin.v2/libs/chrono/build/darwin-4.2.1/release/threading-multi/libboost_chrono-mt.dylib" "bin.v2/libs/thread/build/darwin-4.2.1/release/threading-multi/libboost_thread-mt.dylib" "bin.v2/libs/system/build/darwin-4.2.1/release/threading-multi/libboost_system-mt.dylib"      -headerpad_max_install_names -Wl,-dead_strip -no_dead_strip_inits_and_terms -L/opt/local/lib -Wl,-headerpad_max_install_names  -arch ppc -undefined dynamic_lookup

So this is fixed and the build process continues…

[ballapete (Peter "Pete" Dyballa)]

Replying to Peter_Dyballa@…:

So this is fixed and the build process continues…

Since ten minutes port is installing the boost upgrade…

[ryandesign (Ryan Carsten Schmidt)]

The fix worked for me, and upstream committed it, so I've committed it to MacPorts as well in r140820.

[ryandesign (Ryan Carsten Schmidt)]

Has duplicate #49088.