Re: bjam quits on failure to found a target?

From: David Abrahams (gclbb-jamboost_at_[hidden])
Date: 2003-05-05 07:05:36

Aleksey Gurtovoy <alexy_at_[hidden]> writes:

> Attached is a yesterday night's bjam log produced as the result of our
> nightly Boost regression test run on a fresh CVS checkout (the exact command
> was
>
> bjam --dump-tests "-sALL_LOCATE_TARGET=d:\boost4\results"
>>d:\boost4\results\bjam.log 2>&1
> )
>
> As it appears from the log (which is also available from here -
> http://www.meta-comm.com/engineering/resources/bjam.log), the test run was
> unsuccessful because bjam failed to find a dependent target for _one_ of the
> libraries and just stopped the whole process! That's definitely couldn't be
> the right behavior, could it? 

No.

> We are trying hard to setup a reliable regression system which
> regularly would require no human intervention, and everything being
> broken on an erroneous check-in doesn't really fit into the scheme.

No, it doesn't.

> It would be appreciated if somebody clarified the situation.

Could you add -d+5 at the beginning of the command-line and post the
last hundred lines or so of (or a link to) the resulting bjam.log?

Thanks,
-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com