Re: [Boost-users] boost::thread::join causes assertion (pthreads)

From: Matthew Peltzer (goochrules_at_[hidden])
Date: 2005-03-24 19:41:58

On Thu, 24 Mar 2005 17:53:49 +0000, Ben Hutchings
<ben.hutchings_at_[hidden]> wrote:
> Matthew Peltzer wrote:
> >>Well, if anyone would care to compile and run the following program on
> >>Mac OS X using pthreads, we can at least tell whether there is a
> >>consistent failure:
> >
> >
> > Here's what it looks it looks like with boost1_31 from fink on OS X 10.3.8:
> >
> >
> >>g++ -v
> >
> > Reading specs from /usr/libexec/gcc/darwin/ppc/3.3/specs
> > Thread model: posix
> > gcc version 3.3 20030304 (Apple Computer, Inc. build 1671)
> >
> >>g++ -I/sw/include/boost-1_31 -L/sw/lib -lboost_thread-1_31 -o thread thread.cpp
> 
> It's not clear to me from this whether you linked to a debug version of
> the thread library.  In a non-debug version the assertions are disabled,
> so that wouldn't provide a useful test.

one more time then:

 > g++ -Wall -O0 -DDEBUG  -I/sw/include/boost-1_31 -L/sw/lib
-lboost_thread-d-1_31 -o thread thread.cpp
In file included from
/sw/include/boost-1_31/boost/type_traits/is_arithmetic.hpp:13,
                 from
/sw/include/boost-1_31/boost/type_traits/arithmetic_traits.hpp:14,
                 from
/sw/include/boost-1_31/boost/function/function_base.hpp:19,
                 from
/sw/include/boost-1_31/boost/function/detail/prologue.hpp:16,
                 from /sw/include/boost-1_31/boost/function.hpp:22,
                 from /sw/include/boost-1_31/boost/thread/thread.hpp:20,
                 from thread.cpp:2:
/sw/include/boost-1_31/boost/type_traits/is_float.hpp:21: warning: use of `long 
   double' type; its size may change in a future release
/sw/include/boost-1_31/boost/type_traits/is_float.hpp:21: warning: (Long double 
   usage is reported only once for each file.
/sw/include/boost-1_31/boost/type_traits/is_float.hpp:21: warning: To disable 
   this warning, use -Wno-long-double.)
 > ./thread
Joining exited thread
Joining running thread
 > 

> 
> "Thread model: posix" seems to imply that g++ will generate code
> suitable for pthreads, though most ports require a command-line option
> to ensure that it generates thread-safe code and uses thread-safe
> libraries.  Perhaps the Darwin port always operates in thread-safe mode.

I don't have any answers here.

 
-- Matthew Peltzer
-- goochrules_at_[hidden]