A new trivial kdebase upload needed.
Hi, Sorry that I'm back here whining again, but I don't think we can afford to keep this new serious kdm.postinst bug [1] in the packages for a few months as before. Chris, what are your plans on this ? When do you plan to do a new upload for this ? I can understand if you wait for some more patches to be applied first in order to not make too much uploads of the large kdebase package, but I don't think we can afford to wait more than say a week or so. cheers domi Footnotes: [1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=224548
kdenonbeta/kdedebian/kapture
CVS commit by mornfall: - add TODO - add PkgExistGrouper (filters out "ghost" (no-available-version) packages) - make dealing with multiple versions saner - add version combobox to PkgDetailsCommon - make it show data for selected version - make PkgDetailsCommon use grid layout (that's what it was made for) - reorder PkgGrouperFactory constructor args and make it propagate it's parameters to child if they aren't set explicitly (eliminating the need to pass the params to all the chain members again and again) - make PkgList show current and installed versions NOTE: The code is becoming human-parseable slowly... Still, i have made up some design in my mind that's not incarnated in code yet. Stay tuned: more to come. ATODO 1.1 M +1 -0 kapture/kapture_part.cpp 1.4 M +26 -0 kapture/pkgdetails.cpp 1.2 M +9 -2 kapture/pkgdetails.h 1.2 M +58 -4 kapture/pkgdetailscommon.cpp 1.2 M +9 -2 kapture/pkgdetailscommon.h 1.2 M +287 -335 kapture/pkgdetailscommonui.ui 1.2 M +37 -2 kapture/pkggrouper.cpp 1.6 M +15 -4 kapture/pkggrouper.h 1.6 M +21 -9 kapture/pkglist.cpp 1.5 M +7 -1 kapture/pkglist.h 1.4 M +1 -0 kapture/pkgsubtree.h 1.4 M +1 -0 kapture/pkgtreeitem.h 1.5 M +1 -1 kapture/pkgtreenode.h 1.5 M +12 -2 kapture/pkgview.cpp 1.3 M +2 -0 kapture/pkgview.h 1.2 M +1 -1 kapture/pkgviewui.ui 1.3
Re: A new trivial kdebase upload needed.
On Mon, Dec 22, 2003 at 11:37:06AM +0100, Dominique Devriese wrote: > > Hi, > > Sorry that I'm back here whining again, but I don't think we can > afford to keep this new serious kdm.postinst bug [1] in the packages > for a few months as before. > > Chris, what are your plans on this ? When do you plan to do a new > upload for this ? I can understand if you wait for some more patches > to be applied first in order to not make too much uploads of the large > kdebase package, but I don't think we can afford to wait more than say > a week or so. There appears to be another RC bug now that needs to be fixed as well, which I unfortunately don't know how to fix. It causes kdebase to fail to build on both mips/mipsel. http://buildd.debian.org/fetch.php?&pkg=kdebase&ver=4%3A3.1.4-1&arch=mips&stamp=1071907102&file=log&as=raw http://buildd.debian.org/fetch.php?&pkg=kdebase&ver=4%3A3.1.4-1&arch=mipsel&stamp=1071915371&file=log&as=raw The problem file is kdebase/kcontrol/info/memory_linux.cpp In file included from ../../../kcontrol/info/memory.cpp:433, from kcm_info_la.all_cpp.cpp:3: ../../../kcontrol/info/memory_linux.cpp: In member function `void KMemoryWidget::update()': ../../../kcontrol/info/memory_linux.cpp:14: error: `SYS_sysinfo' undeclared (first use this function) ../../../kcontrol/info/memory_linux.cpp:14: error: (Each undeclared identifier is reported only once for each function it appears in.) make[4]: *** [kcm_info_la.all_cpp.lo] Error 1 /usr/include/bits/syscall.h defines SYS_sysinfo on most archs though so is this a bug in glibc for mips/mipsel? Or should the code be using sys/sysinfo.h instead of bits/syscall.h? Andrew Suffield mentioned it probably should be using "sysinfo(&info)" instead of using syscall. Does that sound correct to you? IMHO we should fix this problem in the next upload as well. Thanks, Chris signature.asc Description: Digital signature
Re: A new trivial kdebase upload needed.
Chris Cheney writes:
> There appears to be another RC bug now that needs to be fixed as
> well, which I unfortunately don't know how to fix. It causes kdebase
> to fail to build on both mips/mipsel.
>
Apparently there's a discussion on #kde-devel going on about this.
DanielS is there, fixing the syscall invocations in CVS to something
more portable. I took his patch for kdebase ( there were some
problems in kdeutils too, but those are less urgent, I guess ), and
built it into a patch against the debian/* dirs. Attaching below.
I have tested the "debian/rules configure", and it still works, the
new patch in debian/patches applies cleanly. I have not tested the
entire compilation, but I assume DanielS has done this properly.
> IMHO we should fix this problem in the next upload as well.
Agreed. It's just that before the last upload was made, there were
some 2 months that kdebase had major RC bugs, and I don't think we can
afford that again, especially at this stage in the Debian release
cycle.
Thanks for your work.
cheers
domi
diff -u -rNubB kdebase-3.1.4/debian/patches/07_syscall.diff kdebase-3.1.4.mod/debian/patches/07_syscall.diff
--- kdebase-3.1.4/debian/patches/07_syscall.diff 1970-01-01 01:00:00.0 +0100
+++ kdebase-3.1.4.mod/debian/patches/07_syscall.diff 2003-12-23 00:55:47.0 +0100
@@ -0,0 +1,29 @@
+Index: kcontrol/info/memory_linux.cpp
+===
+RCS file: /home/kde/kdebase/kcontrol/info/memory_linux.cpp,v
+retrieving revision 1.5
+retrieving revision 1.6
+diff -u -u -b -B -r1.5 -r1.6
+--- kdebase/kcontrol/info/memory_linux.cpp 27 Jan 2002 17:34:24 - 1.5
kdebase.mod/kcontrol/info/memory_linux.cpp 22 Dec 2003 23:22:10 - 1.6
+@@ -1,17 +1,17 @@
+-#include
++#include
+ #include
+ #include
+ #include
+ #include
+
+-/* $Id: memory_linux.cpp,v 1.5 2002/01/27 17:34:24 adrian Exp $ */
++/* $Id: memory_linux.cpp,v 1.6 2003/12/22 23:22:10 dstone Exp $ */
+
+ void KMemoryWidget::update()
+ {
+ struct sysinfo info;
+ int shift_val;
+
+- syscall(SYS_sysinfo, &info); /* Get Information from system... */
++ sysinfo(&info); /* Get Information from system... */
+
+ /* try to fix the change, introduced with kernel 2.3.25, which
+ now counts the memory-information in pages (not bytes anymore) */
