Boost Testing :
From: Martin Wille (mw8329_at_[hidden])
Date: 2006-11-15 14:29:30
Juergen Hunold wrote:
>> I hope there's a compelling reason for statically linking.
> The only reason is to be consistant on all platforms. The use of shared
> libraries with msvc requires special support. This support is not
> available for serialisation, so we link static for _all_ platform now
> in order to get (nearly) the same errors on all platforms.
> In theory, I could try to add a small static library to those libraries
> containing the "main" function. That would increase size slightly.
> But that has to be tested with msvc.
> So I'll wait and see what Volodya does...
I was suspecting something like that.
Frankly, I'm unhappy that a shortcoming that originally affects a single
platform was turned into an inferior solution for all platforms.
I'm not so much worried about HD space (I'm sure we're going to have
working removal code soon; Volodya checked something in a moment ago),
but about time. Static linking takes quite a bit a longer than dynamic
linking. This will slow down the tests.
If there are different errors, depending on whether you link statically
or dynamically, then there are probably problems that need investigation
(I'm not talking about platforms for which dynamic linking is currently
not well supported). With the exception of Boost.Python, there shouldn't
be any complicated linking setups in Boost, AFAIK.
Wouldn't it be possible to add some variable to the build system that
indicates the preferred linking style for the tests on a given platform?
Send instant messages to your online friends http://au.messenger.yahoo.com