Boost Testing :
From: AlisdairM (alisdair.meredith_at_[hidden])
Date: 2006-11-20 16:01:09
John Maddock wrote:
> OK, I hadn't realised it was a release build, I can reproduce now,
> looks like a known linker bug:
> http://qc.borland.com/wc/qcmain.aspx?d=32020 that occurs when the
> project exceeds a certain size. I guess I'll just have to mark them
> up as expected :-(
I'm not sure that is the right answer - at least not for Boost 1.34.
I am concerned that JustSoftwareSolutions are not so much reporting
regressions, as running tests in new configurations - release builds.
This is an incredibly valuable contribution, but nailing these problems
is new work, not fixing regressions. Given that we really want to ship
1.34 sooner rather than later, I would suggest this test runner focus
back on the mainline, and we pay attention to release-configuration
issues for 1.35.
In particular, I am wondering if we can mark up expected failures
separate for release vs. debug builds, and now is not the time to be
working out answers to such questions!