Cal, Pawel, Baho, all,
For the arch master build script I have the script update the tree and then
create tarballs on the fly. (easier that mount --bind of local git tree into
chroot to use the git tree directly). One of the things I had to do was to
automate the update of the md5sum in the PKGBUILD. I have come up with a rather
wonky 2-function approach that I want to get comment on. May be OK as it (it
works fine), but I welcome thoughts for improvement. The only limitation is that
the tarball must be the first md5sum listed in the md5sum array. The functions I
use are:
## function getsum(): get the new md5sum (getsum filename)
getsum() {
[[ -r "$1" ]] || { echo "ERROR: file not found '$1' in function getsum";
return 1; }
_tmp=$(md5sum "$1")
echo "${_tmp//\ *}"
}
## function updtsum(): update the md5sum in pkgbuild (updtsum file newmd5sum)
updtsum() {
sed -i -e "s/md5sums=('.*'/md5sums=('$2'/" "$1"
}
Then they are called in the script as follows:
## update the md5sum
_newsum=$(getsum "${tgzdir}/${tgzfn}")
echo " updating PKGBUILD md5sum -> $_newsum"
updtsum "${pbpkgd}/${pbpkg}/PKGBUILD" $_newsum
Any areas you see that could be improved? The script isn't ready for
distribution yet, but it is getting close.
--
David C. Rankin, J.D.,P.E.
Tim, Calvin,
In the tqtinterface build setup, I find a tqtpc-location.patch that isn't
being utilized. It is:
--- dependencies/tqtinterface/CMakeLists.txt 2011-03-28 00:12:24.000000000 +0200
+++ dependencies/tqtinterface/CMakeLists.txt.new 2012-01-29
12:15:16.000000000 +0100
@@ -35,7 +35,7 @@
tde_setup_install_path( BIN_INSTALL_DIR "${EXEC_INSTALL_PREFIX}/bin" )
tde_setup_install_path( LIB_INSTALL_DIR
"${EXEC_INSTALL_PREFIX}/lib${LIB_SUFFIX}" )
tde_setup_install_path( INCLUDE_INSTALL_DIR
"${CMAKE_INSTALL_PREFIX}/include/tqt" )
-tde_setup_install_path( PKGCONFIG_INSTALL_DIR "/usr/lib${LIB_SUFFIX}/pkgconfig" )
+tde_setup_install_path( PKGCONFIG_INSTALL_DIR
"${EXEC_INSTALL_PREFIX}/lib${LIB_SUFFIX}/pkgconfig" )
##### tqtinterface sources ######################
It isn't applied in the latest GIT tree -- do we need it?
--
David C. Rankin, J.D.,P.E.
Cal, Pawel, Baho, all,
To keep things straight with PKGBUILDs and source files (including PKGBUILD
and patches, etc..) between those that build with Qt3 and those utilizing TQt3,
I have simply made a name change:
All Qt3 based packages and sources will be trinity-xxx. Eg:
trinity-tqtinterface
trinity-arts
All TQt3 based packages will be tde-xxx. Eg:
tde-tqtinterface
tde-arts
...and so on. Let me know your thoughts on this. I thought this made more sense
than monkeying with the $pkgver or other variables.
--
David C. Rankin, J.D.,P.E.
This message is from tdelibs/kinit/kinit.cpp. They appear in the xession log.
I don't mind the messages and I can see from the source code they are intended to be debugging aids.
But they lack information. How do I add the process name to the message?
Like this:
PID 10804 (process name) terminated.
With that information the message might remain useful only for debugging, but at least the message appears more useful and informational to users.
Darrell
I'm trying to download the git files as an anonymous user.
When it starts downloading, everything goes fine for the first 43 megs
and 50 files. All the files are put in the correct place in my build
tree--but then it stops downloading and displays:
-----------------
Cloning into experimental...
Password:
-----------------
When pressed enter, it started downloading again. But from then
on--every minute or so it would stop downloading and request a
password.
I Googled and read the man page, several tutorials and HowTo articles
but found nothing that seemed to apply to this problem.
I am using a fresh Slackware 13.37 install less KDE4 and my version of
git is 1.7.4.4.
I used this get command from the Trinity Project GIT Information page first:
git clone --recursive http://scm.trinitydesktop.org/scm/git/tde $GIT_TREE
Then I tried this instead, since some articles used this syntax:
git clone http://scm.trinitydesktop.org/scm/git/tde --recursive $GIT_TREE
The results were identical in both cases.
($GIT_TREE is the variable for the name of the directory where I save
the files.)
Does anyone have any idea what I am doing wrong?
I can't sit there for hours just waiting to press the enter key... I
could rig the script to send enter everytime it sees "Password:" but I
would rather know if there is something wrong before I start trying to
build a workaround kludge.
Keith
Continuing to shed light on my issue.
The same appeared on second host. Both have Ubuntu 'Precise' on it.
GCC version is different: 4.6.2 vs 4.5.3. cmake version: 2.8.5 vs
2.8.7. Looks like these two are not the cause of the problem. I should
really look at uic sources.
Will remind what's going on:
[ 18%] Building CXX object
kdeui/CMakeFiles/kdeui-shared.dir/kshortcutdialog_simple.cpp.o
/home/midenok/src/kde/tdelibs/build.debug/kdeui/kshortcutdialog_simple.cpp:
In constructor ‘KShortcutDialogSimple::KShortcutDialogSimple(QWidget*,
const char*)’:
/home/midenok/src/kde/tdelibs/build.debug/kdeui/kshortcutdialog_simple.cpp:31:50:
error: invalid use of incomplete type ‘struct QLabel’
Adding '#include <qlabel.h>' to kshortcutdialog_simple.cpp fixes it.
But moves straight to similar error:
[ 10%] Building CXX object
tdeui/CMakeFiles/tdeui-shared.dir/kshortcutdialog_advanced.cpp.o
/home/midenok/src/kde/tdelibs/build.debug/tdeui/kshortcutdialog_advanced.cpp:
In constructor ‘KShortcutDialogAdvanced::KShortcutDialogAdvanced(QWidget*,
const char*)’:
/home/midenok/src/kde/tdelibs/build.debug/tdeui/kshortcutdialog_advanced.cpp:33:58:
error: invalid use of incomplete type ‘struct QButtonGroup’
Looking at kshortcutdialog_simple.ui shows:
<widget class="TQLabel" row="0" column="0">
I suppose uic must add any q*.h include to any TQ* class it finds in .ui?
Hello
I'm going to finish the next update to 3.5.13 kdebase. I added the following:
Fix kdm_greet high CPU usage in SAK dialog and without SAK - perhaps the final
solution of 100% CPU load in KDM.
Upon recommendation by David Hare, I update the startkde script (based on
commit 39fcb49a).
Based on commit d409abdb from KDE 4.6, I rewrote the patch to hostname not
displayed in the window title if it is the FQDN of localhost. (It does
LibreOffice for example.)
Does anyone have more ideas before they will go to compile the package in the
PPA?
Slávek
--
Different system. I'm seeing this message in my xsession log. The message occurs after I see messages about autostart apps and in the messages where kcompmgr seems to be starting. Messages immediately following:
No composite extension
trying '/root/.xcompmgrrc' as configfile
finished parsing the config file
Which probably indicates the error is associated with kcompmgr.
The message itself is from tqt3 qlayout.cpp:1227:
qWarning( "TQLayout: Adding %s/%s (child of %s/%s) to layout for "
Any ideas how to resolve?
Darrell
I'm seeing this message in my xsession log. I added the [tdmctl] in a patch I have not yet submitted to clarify the source of the message.
==========
Bug No. 1:
==========
I am starting X/TDE from the command line through the startx script. TDM is not running and should not be in the big picture anywhere.
In the code I found only two apps that call tdmctl: the TDM greeter and tsak. I rebuilt tdebase with -DBUILD_TSAK=OFF. The message then disappeared from my xsession logs. Culprit found but next needed a coding solution.
In tdebase/tdmlib/tdmtsak.cpp there are two calls to tdmctl. Seems this message could be handled more gracefully to avoid the mysterious xsession messages. Patching "2>/dev/null" in the source code and rebuilding without -DBUILD_TSAK=OFF again eliminated the messages.
But is that what we want? I don't know the specific purpose for the error message. Possibly the message is useful, but seems a more graceful approach is to test whether TDM is running before trying to execute the tdmctl command.
I presume tsak remains useful when a session is started from the command line for when a user locks the session. If so then again, there should be a test before each call whether TDM is running.
The error does not appear when TDM is running. I think a simple TDM trap test is all that is needed.
In C++ how do I perform the equivalent of ps ax | grep tdm?
Should tsak be running when TDM is not?
==========
Bug No. 2:
==========
Both calls to tdmctl in tdmtsak use the format "tdmctl list" yet there is no "list" option in tdmctl --help. What is the "list" parameter supposed to do?
==========
Bug No. 3:
==========
When I built tdebase without tsak support (-DBUILD_TSAK=OFF), and in my tdmrc I have UseSAK=false, TDM starts with the dialog to press Ctrl-Alt-Del. That should not happen.
I will submit a bug report but I'd appreciate some conversation about the topic before filing to make sure I'm not missing anything obvious.
Darrell