From: Jens Seidel (jensseidel_at_[hidden])
Date: 2008-03-27 17:31:12
On Tue, Mar 25, 2008 at 03:57:43PM +0000, Phil Endecott wrote:
> Jens Seidel wrote:
> > std::system("true");
> > // leads to "Test setup error: child has exited; pid: 1001; uid: 30540; exit value: 0"
>
> > Any idea?
>
> I know basically nothing about Boost.Test, but I won't let that stop me
> from guessing that it has a signal handler for SIGCHLD, or conceivably
> a thread blocking in a call to wait(), that detects your subprocess
> terminating. Perhaps it's not expecting your test to start subprocesses?
You're probably right. I didn't know that there exist such a signal
(I know only the most basic ones).
I reported it as bug #1723 since this behaviour is new and some
non-portable cleaning of log files via system("rm file.log") worked in
the past.
Jens