Re: [boost] Re: Regression Tests, Boost.Serialization

From: Martin Wille (mw8329_at_[hidden])
Date: 2004-07-16 13:34:30

David Abrahams wrote:
> Martin Wille writes:
> 
> 
>>Martin Wille wrote:
>>
>>> Any other suggestions? 
>>
>>Would it be viable to modify the build system
>>such that test binaries which ran successfully
>>get stripped after they ran? 
> 
> 
> It might be possible...
> 
> 
>>Having the then unneeded debug symbols removed would help a lot.
> 
> 
> ... but it might be painful.  

For test binaries which ran successfully? What's the use
of debug symbols in a test binary that did not produce
any errors? I'm not thinking of removing debug information
from test binaries which produced errors.

> What are the required commands to strip an executable?
> 

strip filename

Test results should be available within a few hours
(including Serialization. I removed gcc-3.4.0 from
the list of compilers. 3.4.1 is still present).

While tests are running, I'm seeing link errors
for Boost.Serialization:
/boost/head-regression/boost/boost/test/test_tools.hpp:179: undefined 
reference to `pthread_mutex_destroy'

I think my installation of libstdc++ requires
-pthread on the commandline. Is this a Boost.Build
problem, a Boost.Serialization problem or do I just
have to add some options to my toolset definition
for gcc-3.4.1?

Regards,
m