I wrote autotools files for portmidi and created a tarball of it. At the
moment it only compiles for linux. I then created a gub file. You can
see that in gub/specs/portmidi.py after doing a git pull. Once I have
this working for windows also we can remove
portmidi from denemo. I think it is a mess anyway. Is there any reason
to keep it with out code if we get this working reliably?
I am currentan ly uploading denemo-1.0.0.~rc.7-3.linux-x86 that I just
created linking against this libportmidi.la that was created from this.
I tested and confirm that portmidi is working this way in the linux
build. I am using the master branch becuase the linux branch does not
seem to be compiling for now. That is ok. The only change I need it is
to patch denemo/src/prefops.c to force denemo to use the included
lilypond. Then its just of getting evince to open up pdf's in gtk3 based
systems to get the linux binary in get shape
Jeremiah
On 01/22/13 07:58, Richard Shann wrote:
Jeremiah -
I am making good progress in getting mingw building with portmidi. With
the mingw branch set in denemo.make the portmidi.h header was not being
found, so I modified the sources to include "portmidi/portmidi.h" which
gets them found. I have got to the stage where the libportmidi.a library
is built and the libaudiobackend.a is built, and all the needed symbols
(e.g. _Pt_Start) are present for portmidi use. The final hurdle is that
libportmidi.a is not being linked to (libaudiobackend.a is being linked
in, but not libportmidi.a which remains in ../portmidi relative to the
build directory).
I think I spotted one mistake in your gub, you have
--with-static-portmidi in denemo.py where I think it should be
--enable-static-portmidi
e.g.
my denemo.py now has:
class Denemo__mingw__console (Denemo__mingw__windows):
#patches = ['portmidi-denemo-test.patch']
configure_flags = (Denemo__mingw__windows.configure_flags
+ ' --disable-binreloc'
+ ' --disable-portmidi'
+ ' --enable-static-portmidi')
configure_variables = (Denemo.configure_variables
+ ' CFLAGS="-D_HAVE_PORTMIDI_ -I
%(system_prefix)s/include/evince/2.30" '
+ ' LDFLAGS="-L%(system_prefix)s/lib -levview
-levdocument" ')
where additionally I specified the -D_HAVE_PORTMIDI_ in the configure
vars, though I don't know that this is needed.
One of the problems preventing me quickly testing for things like that
is that I can only force a rebuild by doing a git push on the mingw
branch. (As you mentioned gub will not rebuild if it thinks the source
is unchanged). It should be possible to change some checksum value to
force a rebuild - you haven't found out where that is by any chance. I
am nervous that I will trigger a complete rebuild of everthing...
Richard
_______________________________________________
Denemo-devel mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/denemo-devel
