On Jul 7, 2009, at 05:14 , Peter Soetens wrote:

> 2009/7/2 orocos <orocos [..] ...>:
>> Hi,
>> is there any way to speed up the build time of applications
>> including orocos headers?
>>
>> I did some tests precompiling the orocos headers with GCC 4.4 and,
>> as described at http://lists.mech.kuleuven.be/pipermail/orocos-dev/2004-August/000295.html
>> , compilation actually gets slower.
>
> The ways we optimize are by using ccache and setting
> PATH=/usr/lib/ccache:$PATH and also compiling with 'make -j2' (or
> higher). Next you can use distcc to distribute to other hosts which
> have the same compiler installed. For combining ccache and distcc,
> you need to set the environment variable CCACHE_PREFIX=distcc as well.
> It's alittle setup time but it pays off *very* quickly. We're also
> considering constantly how we can improve the RTT headers themselves
> such that they compile faster. Compiling with -O0 (or -O1) also
> dramatically improves speed with recent gcc compilers.

Have you thought about implementing any of Lakos' recommendations from
"Large Scale C++ Software Design"? Plenty of speed-ups offered there
that don't require additional tools. The book also provides tools to
analyse existing systems to see where speedups might be possible.
Stephen