I just added 'optimize=highest' to the SCons build. This causes XORP to be built with the -O3 flag passed to gcc. There is negligible difference in binary size vs optimize=yes (gcc -O1, the default). I haven't measured performance. thanks, BMS