Re: [boost] Windows and apache stdcxx

No, you misunderstood. I _did_ build with consistent wchar_t option (off in that case). But boost won't build the date time library without native wchar_t. I tried with and without stdcxx library on windows with vc7.1, it never worked. I needed to turn wchar_t on for it to work (with <native-wchar_t> or so....), there seems to be kind of a boost bug ;o) but i'm not sure about that. On vs8 this is not a problem, since wchar_t is native by default. On vc7.1 i don't want to rebuild everything, better: i don't want to change our build mechanisms, since those are many, and not too easy to change ;o) Cheers, Markus -----Original Message----- From: boost-bounces@lists.boost.org [mailto:boost-bounces@lists.boost.org] On Behalf Of John Maddock Sent: Monday, September 25, 2006 11:02 AM To: boost@lists.boost.org Subject: Re: [boost] Windows and apache stdcxx Duft Markus wrote:
Hey!
The thing is, that the stdcxx reall should behave as if it was part of
the compiler, since it's integrated into my wgcc program, so nobody should have to do anything special to use it (no need to define or include anything, it just replaces the MS libc++) The thing is, that i
allready got really far with a little patching ;o) the only thing is, that date_time wouldn't build, because of some weird errors in lexical_cast.hpp where some operators are not found. This has to do with not useing /Zc:wchar_t. I don't understand why this breaks compilation, but i didn't want to compile stdcxx with Zc:wchar_t. Since we have an automated build system it would be a great deal of work to change the wchar_t behaviour in all builds. But at the moment i disabled stdcxx, because it didn't work out that good, so we're now waiting for things to get better ;o)
Oh right, if you build the std lib without /Zc:wchar_t and build boost with it, then that absolutely will lead to linker errors at the very least. Try building with consistent /Zc:wchar_t options and that should fix things (hopefully!). John. _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Duft Markus wrote:
No, you misunderstood. I _did_ build with consistent wchar_t option (off in that case). But boost won't build the date time library without native wchar_t.
Well, that's fairly hard for me to imagine based on what is in the lib. The tests are a different story...many platforms have issues with correct support for wchar and hence fail a variety of date-time tests designed to exercise wchar support. Perhaps you need to post the failure details....or did I just miss them? The other thing I'll mention is that the platforms that use the Rogue Wave library, in general, haven't been the best for Boost support in general. So I wouldn't be surprised at all if there are plenty of boost compatibility issues with apache stdcxx. As an example, in date-time I had to add some code for facets that returned id's to make things work with the RW library: //time_facet.hpp #if defined (__SUNPRO_CC) && defined (_RWSTD_VER) std::locale::id& __get_id (void) const { return id; } #endif I don't know if this is still needed, but if it is you won't be getting it because the config macros won't select it. If you search the boost source on RWSTD you'll find other workarounds. Jeff
participants (2)
-
Duft Markus
-
Jeff Garland