Re: [Mingw-cross-env-list] SDL_Pango in mingw-cross-env

[David Bruce]

Hi Volker,

> Although your solution seems to be less "clean", I like it more
> because it works equally well and requires just a minimum of changes
> to the original sources.
>
> So I included your patch. Thanks for your contribution!
>
> http://hg.savannah.gnu.org/hgweb/mingw-cross-env/rev/87ceb4929ce9

Glad to hear that it is in mingw-cross-env, and thanks for the
feedback on the shortcomings of my diff.  One thing I am not quite
clear on, however, is why the "multiple definitions" error doesn't
show up when SDL_Pango is used in linux, where it is dynamically
linked.  Would this sort of bug (and I am pretty sure that having
ordinary variable definitions in a header file is a bug) only show up
with static linking?

Also, there are still several issues with my mingw-cross-env based
crossbuilds of tuxmath and tuxtype, some of which likely have nothing
to do with mingw-cross-env per se, but others which might reflect
things that need your attention.

With respect to SDL_Pango, the programs now build successfully and run
without crashing, but they do not actually display any text when I run
them on my Windows test machine.  Has anyone on this list used
mingw-cross-env's SDL_Pango with other programs?  Since this may well
not be an issue with mingw-cross-env itself, I would ordinarily ask
the upstream developers, but SDL_Pango doesn't seem to have any
upstream devs.

Along those lines, I see that there are some test *.c files in
mingw-cross-env for some of the SDL libs.  What is the recommended way
to compile these for testing purposes?  Should the test programs do
something like display "Hello World" or make a sound, rather than just
see if the initialization function works and then exit?

Briefly, the other "issues" are the following, although it probably
makes sense to address them in separate threads:
1. AC_CHECK_LIB and PKG_CHECK_MODULES both fail to find and link
SDL_ttf, unless SDL_Pango is linked first, apparently because the
configure process doesn't link in freetype when SDL_ttf is being
checked for.  I can provide details in a separate thread.
2. The check for SDL_net does succeed, although the check for ws2_32
fails - have not yet checked to see if the resultant binary can use
SDL_net successfully at run time (details to follow).
3. The SDL_mixer unexpected exit is still unresolved - I asked on the
SDL list and got some help, but didn't get the problem fixed - maybe
time to ask again.

Anyway, thanks very much for your help.

David Bruce