[Boost-testing] process_jam_log and boost.test

From: Vladimir Prus (ghost_at_[hidden])
Date: 2007-10-25 03:39:19

Martin Wille reported on IRC that process_jam_log fails to build on trunk
due to boost.test:

[11:28] <_m_> bin/gcc-3.4.6/release/process_jam_log.o: In function `execution_monitor_attaching_signal_handler':
[11:28] <_m_> process_jam_log.cpp:(.text+0x97e): undefined reference to `boost::debug::attach_debugger(bool)

Looking at process_jam_log, it uses cpp_main instead of main, and the only point of that, IIUC,
is to get slightly nicer output on errors. Does this benefit really outweights the cost of
extra dependency? Really, if process_jam_log decides to crash with SIGSEGV, then it does not
matter if it just crashes or prints "SIGSEGV" on stdout -- in both cases, only somebody
equipped with a debugger can fix the problem.

- Volodya