[Omniprint-developer] Re: [Gs-devel] Omni and makefile changes

[rillian <ri...@te...>]

On Wed, 23 Jan 2002, our fearless leader wrote:

 The Omni.patch.090501.clean patch is not suitable as is for a
Ghostscript release. Basically, the patches have to be localized
to the unix-gcc.mak toplevel makefile. The patches you've made to
ugcclib.mak, and unixlink.mak would rather dramatically break the
build for other platforms.

Here's a basic stab at the integration. It's not finished; in particular the load of libomni.so fails due to missing symbols, so the build isn't quite right, though that could certainly be on the Omni side, rather than with ghostscript.

This patch updates gomni.c to the version from Omni-0.5.1 (the 090501.clean patch) and adds some rough support for building it to the makefiles. The tricky bit is that Omni now depends on glib, building against which pretty much requires glib-config calls on the command line. One option would be to add -compileliteral, and -linkliteral .dev args to genconf. What we do here is use a bit of sed to strip out the linker command options and pass that from the toplevel makefile so genconf can put them back in. We cheat on the cflags and pass them directly, since there's no similar mechanism of -I lines. $(I) and friends should be handled by glib-config on any platform we'll be able to build this on anyway, so I think we lose no portability there.

The patch is against GS_6_5, but is intended for all branches. It also does some rearrangement in contrib.mak. Sorry for the noise.

TODO:
  port the device documentation from Omni.readme.1st
  fix linker options

FWIW,
 -r