#19 allow t disable USB support

[Waldemar Brodkorb]

Hi,
the Buildroot project have some build errors regarding gutenprint.
It would be nice if the USB backend could be disabled via configure.

~~~~
Allow to disable USB backend

Signed-off-by: Waldemar Brodkorb wbx@openadk.org

diff -Nur gutenprint-5.2.11.orig/configure.ac gutenprint-5.2.11/configure.ac
--- gutenprint-5.2.11.orig/configure.ac 2016-01-14 04:02:49.000000000 +0100
+++ gutenprint-5.2.11/configure.ac 2016-04-02 20:07:54.649589473 +0200
@@ -442,12 +442,15 @@
dnl [EXTRA_LIBREADLINE_DEPS]=""
dnl fi

+AC_ARG_WITH([libusb], AS_HELP_STRING([--with-libusb], [Build with USB backend support]))
# libusb-1.0 (For CUPS backends)
+AS_IF([test "x$with_libusb" = "xyes"], [
PKG_CHECK_MODULES([LIBUSB], [libusb-1.0],
[HAVE_LIBUSB=yes
BUILD_LIBUSB_BACKENDS=yes],
[HAVE_LIBUSB=no
BUILD_LIBUSB_BACKENDS=no])
+])

AC_PATH_PROGS([GIMPTOOL2_CHECK], [gimptool-2.0 gimptool])
if test -z "${GIMPTOOL2_CHECK}" ; then

---

[Solomon Peachy]

Can you paste the actual build problems you're seeing? Because the configure script shouldn't barf if libusb is not present (cleanly disabling the backends), but if libusb is present and gutenprint is failing to build for some reason, we should fix the actual build problem rather than simply defaulting the libusb stuff to be off.

(I'd like to see the configure invocation used, its output, plus the build output where things fail)

[Solomon Peachy]

Inspecting the configure.ac file I see that it's technically possible for BUILD_LIBUSB_BACKENDS to be enabled even if BUILD_CUPS is not. That may be the root cause of the problem.

[Waldemar Brodkorb]

Yes, indeed, that is the real reason.
See here for the error log from the autobuilders:
http://autobuild.buildroot.net/results/081b3be918ac1eaa8cfbc5919e00bc1ea267c1df/
And my second patch, which should be ignored:
http://buildroot-busybox.2317881.n4.nabble.com/PATCH-v2-gutenprint-fix-host-build-td131712.html

Peter got the same results as you:
http://buildroot-busybox.2317881.n4.nabble.com/PATCH-gutenprint-fix-host-build-when-libusb-found-td130902.html

[Solomon Peachy]

I committed this patch, which should solve the immediate problem:

https://sourceforge.net/p/gimp-print/source/ci/fc1713589fed645f2c8002c3f66dfb69af314716/

Perhaps more troubling is that according to the buildroot config, CUPS is enabled but gutenprint is not picking it up. it looks like we're keying off of 'cups-config' being able to be invoked (as per upstream guidelines) but that needs to be on the path.

Without CUPS, Gutenprint is much less useful. Does anything else in buildroot depend on it?

(Unfortunately, it appears as if cups isn't terribly cross-compilation friendly..)

[Waldemar Brodkorb]

Thanks, this patch is working in our case.
Gutenprint for the target is with CUPS enabled, just the previous host compile isn't.
It seems in the cross-compile process some parts of gutenprint are required to run on the build host.
I am not really involved in this package creation, I just have seen a broken sparc64 package and
I try to keep autobuild failures for sparc/sparc64 low :)