third set of Hurd fixes, and libOSMesa problem

[Marcus Brinkmann]

Hi,

these are the remaining items I sorted out today. I am actually up to
dh_movefiles now, so building the packages will be easy (dh_movefiles is so
slow...).

The three patches attached (113 - 115) fix a couple of issues the resulting
MANIFEST file revealed. The MANIFEST file after applying 113-115 is
attached, too. This is actually very much like the i386 MANIFEST file,
except the fbdevhw and v4l drivers, the DRI stuff, and libOSMesa.

With libOSMesa, a problem arises: The Hurd does not have the DRI/DRM kernel
support Linux has, so libOSMesa is not build, and the packages should not be
build at all. Strictly spoken, the packages libosmesa* are
Architecture: i386, sparc, alpha, powerpc, etc... not listing hurd-i386

I have a proposal to make it possible to specify:
Architecture: linux-any
to give the same effect, but no response to the posting on debian-dpkg yet.

Branden, this can also be worked around by preprocessing the control file
with cpp (or using two control files etc). This is a bit ugly, but works.
Also acceptable as a kludge would be just to "touch" the missing files, and
have dummy packages, or even providing an empty libosmesa3.files.hurd-i386
file (in this case, the other arches have to use their own *files file, too,
of course).

Thanks,
Marcus

-- 
`Rhubarb is no Egyptian god.' Debian http://www.debian.org [EMAIL PROTECTED]
Marcus Brinkmann              GNU    http://www.gnu.org    [EMAIL PROTECTED]
[EMAIL PROTECTED]
http://www.marcus-brinkmann.de

Two bugs in the gnu.cf file from patch #110:
* Build xf86rush.
* Put kbd/README in /var/lib, not /var/state.

--- xc/config/cf/gnu.cf.unpatched       Wed Oct 25 23:51:20 2000
+++ xc/config/cf/gnu.cf Thu Oct 26 00:36:29 2000
@@ -42,6 +42,11 @@
 #ifndef HasGcc2ForCplusplus
 # define HasGcc2ForCplusplus   YES
 #endif
+
+#ifndef BuildXF86RushExt
+# define BuildXF86RushExt      YES
+#endif
+
 #define GccUsesGas             YES
 #define UseGas                 YES
 #define GnuCpp                 YES
@@ -122,7 +127,7 @@
 #define StandardCppDefines     StandardDefines
 
 #define HasVarRun              YES
-#define VarDbDirectory         /var/state
+#define VarDbDirectory         /var/lib
 
 XCOMM i386Architecture
 #define OptimizedCDebugFlags   DefaultGcc2i386Opt

This patch fixes a bug in the code triggered if BuildXF86DRI is NO.
The vbeSymbol is not defined in this case, so the symbol must go into
the #if.

--- xc/programs/Xserver/hw/xfree86/drivers/r128/ati2_probe.c.prior      Wed Oct 
25 22:04:34 2000
+++ xc/programs/Xserver/hw/xfree86/drivers/r128/ati2_probe.c    Wed Oct 25 
22:04:53 2000
@@ -263,9 +263,9 @@
 #ifdef XF86DRI
                          drmSymbols,
                          driSymbols,
+                         vbeSymbols,
 #endif
                          fbdevHWSymbols,
-                         vbeSymbols,
                          /* ddcsymbols, */
                          /* i2csymbols, */
                          /* shadowSymbols, */

Build Scanpci on the Hurd also.

--- xc/config/cf/xfree86.cf~    Tue Oct 24 20:32:36 2000
+++ xc/config/cf/xfree86.cf     Wed Oct 25 23:07:03 2000
@@ -911,7 +911,7 @@
     (defined(LinuxArchitecture) && !defined(Mc68020Architecture) && \
      !defined(PpcArchitecture)) || \
     defined(i386BsdArchitecture) || defined(LynxOSArchitecture) || \
-    defined(OS2Architecture)
+    defined(OS2Architecture) || defined(GNUMachArchitecture)
 #   define BuildScanpci                YES
 # else
 #   define BuildScanpci                NO

MANIFEST.hurd-i386.gz
Description: Binary data