It seems firefox3 does not have a cups option. Maybe the run-time CUPS support should always be disabled, and let the gtk print dialog determine the appropriate printers ? Yorick Hardy wrote:
I think the problem might be here: http://mxr.mozilla.org/firefox/source/gfx/src/psshared/nsCUPSShim.cpp which seems to load libcups.so.2 whenever it is available. I have not checked under what conditions the file gets compiled, but it appears the most likely candidate for the trouble. Line 82 of http://mxr.mozilla.org/firefox/source/toolkit/library/libxul-rules.mk also explicitly links to cups, but I don't think that is the problem. Are the attached (untested) patches the right way to go ? Bernd Ernesti wrote:On Tue, Jan 20, 2009 at 09:18:51AM +0200, Yorick Hardy wrote:You probably need to edit your ~/.gtkrc-2.0 and add gtk-print-backends="lpr,file" as described herehttp://library.gnome.org/devel/gtk/stable/GtkSettings.html#GtkSettings--gtk-print-backendsI am guessing this is the machine with the cups option enabled ?Do you happen to have a core dump from one of the other machines (and "thread apply all bt") ?See my mail with such a trace, where gtk is not build with cups.The page you mentioned above says that the default gtk-print-backends is "file,cups".Looks like something is not working correctly and the default is still "file,cups" if cupsis not enabled during build. Bernd
-- Kind regards, Yorick Hardy
--- gfx/src/psshared/nsCUPSShim.cpp.orig 2009-01-20 18:02:07.000000000 +0200 +++ gfx/src/psshared/nsCUPSShim.cpp 2009-01-20 18:03:16.000000000 +0200 @@ -59,6 +59,7 @@ PRBool nsCUPSShim::Init() { + return PR_FALSE; mCupsLib = PR_LoadLibrary("libcups.so.2"); if (!mCupsLib) return PR_FALSE;