On Tue, 2011-07-12 at 10:35 +0200, Lubos Lunak wrote: > On Monday 11 of July 2011, Caolán McNamara wrote: > > I reckon we probably have to bite the bullet and drop -D_GLIBCXX_DEBUG > > from solenv ? > > I think we could simply force --without-system-libs (or at least for C++ > libs) when --enable-debug is used and that could do. I don't how much is > _GLIBCXX_DEBUG is worth, but --without-system-libs doesn't look like that big > price to pay.
I don't know if it's worth it. I mean even if we forced on --without-system-libs for --enable-debug we would still probably need to run around adding the moral equivalent of EXTRA_CFLAGS += -D_GLIBCXX_DEBUG to all the other external c++ modules we have to be sure we don't mix stl versions. I don't think we currently link against any unix system-installed c++ libraries which *don't* have internal equivalents, but if we did we'd be back to square one again I guess. > > p.s. There is another hack e.g. include <debug/list> vs include <list> > > which might be work investigating. > > How would that work? The non-debug 3rd-party lib will always use non-debug > version of list and that means we'd have to manually somehow use non-debug > version of list as well when interfacing with the library. That sounds pretty > unfeasible in practice to me. Yeah, I guess we end up with a link-time error instead of a weird run-time error, which is an improvement I suppose, but not a very useful one. C. _______________________________________________ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice