Re: [Boost-build] Timing changes break tests?

From: Vladimir Prus (ghost_at_[hidden])
Date: 2007-12-11 15:44:09

On Tuesday 11 December 2007 22:54:18 Rene Rivera wrote:
> Vladimir Prus wrote:
> > I'm getting this when running Boost.Build tests:
> > 
> > timedata                  :  FAILED
> > failed command {{{
> > "......"
> > }}}
> > reason {{{
> > error returned by bjam
> > }}}
> > STDOUT {{{
> > 
> > }}}
> > STDERR {{{
> > bjam: compile.c:609: argument_error: Assertion `frame->procedure != 0' failed.
> > 
> > }}}
> > 
> > The 'regression' test fails with the similar problem.
> > 
> > Anybody has an ideal what's up?
> 
> I changed the arguments to __TIMING_RULE__ 
> <http://beta.boost.org/doc/tools/jam/jam/language.html#jam.language.variables.builtins.actionrule>. 
> I didn't realize there was a BBv2 test using it.

Well:
1. bjam should not crash, anyway. Do you think you can covert that to a diagnostic
of some kind?
2. I'm actually not sure about where 'timestamp' comes from. And I'm not sure why 'regression'
even tries to use that timing functionality, but to look into the latter, I need to have
bjam produce a nice stacktrace, not a crash.

- Volodya