|
Boost-Build : |
From: Hartmut Kaiser (hartmut.kaiser_at_[hidden])
Date: 2007-10-10 17:51:20
This has been fixed already (by Beman).
Regards Hartmut
> -----Original Message-----
> From: boost-build-bounces_at_[hidden]
> [mailto:boost-build-bounces_at_[hidden]] On Behalf Of
> David Deakins
> Sent: Wednesday, October 10, 2007 3:53 PM
> To: boost-build_at_[hidden]
> Cc: boost-testing_at_[hidden]
> Subject: [Boost-build] Broken Wave build still preventing
> regression tests [Re: Wave build broken]
>
> Hartmut Kaiser wrote:
> >>
> >> Markus Schöpflin wrote:
> >>> Hello,
> >>>
> >>> I just noted the following lines in the bjam.log created by the
> >>> regression
> >>> tests:
> >>>
> >>> Skipping build: <build>no in common properties
> >>> error: Unable to find file or target named
> >>> error: '/boost/thread//boost_date_time'
> >>> error: referred from project at
> >>> error: '../libs/wave/test/build'
> >>>
> >>>
> >> I just realized that the above error blocks regression tests from
> >> running.
> >> I checked in a change to libs/wave/test/build/Jamfile.v2 which
> >> hopefully fixes the problem.
> >>
> >> Markus
> >>
> > Doh! Copy and paste error (I shouldn't do 'quick'
> workarounds late in
> > the evening...)! Thanks Markus!
> >
> > Regards Hartmut
> >
>
> It looks like the change is not completely resolving the
> issue. We are
> still getting an error message here that is preventing all tests from
> running:
>
> error: Unable to find file or target named
> error: '../../date_time//boost_date_time'
> error: referred from project at
> error: '../libs/wave/build'
>
> I have attached a patch for libs/wave/build/Jamfile.v2 that
> appears to
> resolve the problem (at least it does on our test system).
> If someone
> could apply it, that would be appreciated.
>
> In the bigger picture, it looks like this shows a
> vulnerability in the
> test system, since any error of this type within any jamfile in boost
> can prevent all regression tests from being run. Ideally,
> this kind of
> jamfile error would only prevent the associated library from
> running its
> tests (or libraries that are dependent on it). In practice
> though, it
> seems like it would be difficult to get that kind of behavior
> from bjam,
> at least with the way the tests are currently structured.
>
> Thanks,
> -Dave
>
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk