From: Vladimir Prus (ghost_at_[hidden])
Date: 2006-03-15 01:43:57
On Wednesday 15 March 2006 06:14, Gennadiy Rozental wrote:
> "Vladimir Prus" <ghost_at_[hidden]> wrote in message
> > On Saturday 11 March 2006 08:37, Gennadiy Rozental wrote:
> >> This is pretty critical and needs to be addressed before release.
> >> Unless we are going to be using BB2 and it does not have this
> >> problem. But this needs to be stated.
> > Speaking about V2, can you state what the problem is exactly, what
> > flag do you what to be added, why, and if it should be added either
> > for compilation of Boost.Test, or programs that use Boost.Test, or the
> > entire Boost.
> > With that information, I'll be able to do appropriate changes.
> The problem was clearly stated in OP:
> > Hi,
> > It was brought to my attention that msvc toolset (and all derivatives)
> > doesn't allow to specify /EHa option. The Unit Test Framework *have*
> > to be compiled with this option. It seems that older versions on
> > Microsoft compiler deduced the need in async exception by themselves.
> > This is definitely needs to be address before release.
> > Gennadiy
> What else information do you need?
That OP is unclear in the following respects:
1. What does it mean for msvc toolset to "specify /EHa option"? Do you mean
that toolset does not provide a feature that translates to /EHa. You
certainly can use <cxxflags>/EHa at the moment.
2. Why unit test framework has to be compiled with this option, and what
problem will you get if it's not compiled with it.
3. Should that flag be added only for complication of unit test framework, all
Boost.Test, or all programs that use the unit test framework.
In fact, V2 Jamfile has the following line:
which probably has something to do with /EHa, but without further information,
or testing on your end, I can't really say if that's what you want, or not.
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