On Feb 17, 2012, at 4:26 PM, Josh Andler wrote:
> On Feb 17, 2012 4:21 PM, "Krzysztof Kosiński" <tweenk.pl@...> wrote:
> > > It also suffers from not being able to build on win32. We could put
> > > out a 0.48.3.1 with fixes for both issues if someone can figure out
> > > the Autoconf-fu.
> > We could use this:
> > PKG_CHECK_MODULES(NEW_GLIBMM, glibmm-2.4 >= 2.28, new_glibmm=yes,
> > new_glibmm=no)
> > if test "x$new_glibmm" = "xyes"; then
> > CPPFLAGS="-DG_DISABLE_SINGLE_INCLUDES $CPPFLAGS"
> > fi
> If you could commit this to trunk and us have a few people test, if all is happy then I'd say backporting would be the better of the two options.
There are also checks for GTKMM_2_24 and GTKMM_2_22 in configure.ac in trunk.
So if there is time for a revision, it would be helpful to just change "NEW_GLIBMM" to be "GLIBMM_2_28" instead. If not, this at least shows that the structure presented by Krzysztof is solid and a good approach. Trying the combination, it might look like:
PKG_CHECK_MODULES(GLIBMM_2_28, glibmm-2.4 >= 2.28, with_glibmm_2_28=yes, with_glibmm_2_28=no)
if test "x$with_glibmm_2_28" = "xyes"; then
AC_DEFINE(WITH_GLIBMM_2_28,1,[Build with Glibmm 2.28.x])
Again, though, as long as people can get successful builds tested with what Krzysztof listed, we should be good.