Re: [boost] Re: Linking ther serializaton library was OS X Link Problems

From: Alan Gutierrez (alan-boost_at_[hidden])
Date: 2004-12-21 08:50:45

* Kon Lovett <klovett_at_[hidden]> [2004-12-17 11:52]:
> On Dec 16, 2004, at 9:37 PM, Robert Ramey wrote:
> 
> >Kon Lovett wrote:
> >>>bin/boost/libs/test/build/libboost_prg_exec_monitor.dylib/darwin/
> >>>debug/runti
> >>>me-link-static/shared-linkable-true/libboost_prg_exec_monitor-sd
> >>>-1_32.dylib
> >>>>ld: Undefined symbols: cpp_main(int, char**)
> >>>>
> >>>>
> >>>>    darwin-Link-DyLib-action
> >>>>
> >>>>
> >>> <snip>
> >>>
> >>>These error messages indicate that the serialization test programs
> >>>are being
> >>>built but that the boost.test library isn't being linked in.
> >>
> >>This an attempt to create a dynamic shared library w/ a two-level
> >>namespace & undefined symbols (cpp_main is a "call-out" from the
> >>library code). Not allowed. Must state the library (an "umbrella
> >>framework" usually) to  resolve from at dynamic link time.
> >>
> >>I think this can be solved by shifting to a flat-namespace. I will
> >>try.
> >>
> >
> >I'm not sure what this means.  But I can tell you that the
> >serialization library doesn't call anything in the test library.
> >I presumed that an attempt was being made to compile and run some
> >of the test programs.

> Sorry, I wasn't implying that serialization calls into the test
> library. The test library calls into the user's test program. The
> build of the test libraries is failing here. This is not an
> attempt to run.  It doesn't involve serialization, or any other
> library.

> The creation of test library .dylibs is failing for an esoteric
> darwin reason.

    I'm back at this today. Attempting to get a Boost built for use
    with Reece Dunn's Boost GUI code.

    I'm failing with a simple:
    
        bjam "-sTOOLS=darwin" 

    Just an update.

--
Alan Gutierrez - alan_at_[hidden]