|
From: | Fred Kiefer |
Subject: | Re: back x11 configure tests |
Date: | Mon, 14 Mar 2011 14:53:19 +0100 |
User-agent: | Mozilla/5.0 (X11; U; Linux x86_64; de; rv:1.9.2.14) Gecko/20110221 SUSE/3.1.8 Thunderbird/3.1.8 |
Am 13.03.2011 22:27, schrieb Eric Wasylishen:
Hi, I was getting a missing symbol error when I tried to drag and drop. It turns out that back wasn't linked to libXext; the configure test AC_CHECK_LIB(Xext, main) was failing (silently). This is on Ubuntu 10.10. I changed it to: (and did the same for Xt and Xmu) AC_CHECK_LIB(Xext, main) if test "$HAVE_LIBXext" = no; then PKG_CHECK_MODULES([XEXT], [xext]) fi so it uses pkg-config if the AC_CHECK_LIB can't find the library. This fixed the problem for me, but I'm not sure if it's the right thing to do. One thing is that according to the autoconf docs, PKG_CHECK_MODULES will cause a hard failure if it can't find the library. Are Xext, Xt, and Xmu hard dependencies of x11 back?
Shouldn't we use the values that PKG_CHECK_MODULES set in XXX_LIBS and XXX_CFLAGS. Otherwise, if these get set to non-default values the test will state that the libraries are present, but compilation will fail later on.
I am really not sure which of these extensions are strict requirements for GNUstep to build. Most likely we could change our code to work without them as well. In XGServerWindow.m we use XShapeCombineMask in context.c we use XmuLookupStandardColormap. We also use XShmQueryVersion but these are already protected with checks.
I would be surprised to learn that we actually use Xt, no idea why we are checking for it.
[Prev in Thread] | Current Thread | [Next in Thread] |