> Are any processes using 100% CPU while you are
> waiting? If so, which
> program is using the CPU?
>
> I didn't write the pykdeextensions/python-kde3 package; in
> fact I have
> next to zero Python programming ability, so that is a
> package I don't
> touch unless something is seriously wrong.
I downloaded 3.16.7 and tried compiling that rather than from SVN. Same results. Same stall point.
Top showed the cc1plus compiler hogging 99% CPU time and 8% memory.
This time I waited about 15 minutes and then the build continued.
Possibly then the svn version is okay too.
After continuing to compile --- for only a second --- there was another long stall. The pause was with the kdeui directory.
According to the build instructions (http://www.riverbankcomputing.co.uk/static/Docs/PyKDE3/install.html), compiling should take 15 minutes to one hour.
I waited an hour. Nothing.
I decided to force non-concatenation with the -i switch.
At least the -i switch is verbose and I could see things happening. That mode also was easier on the CPU.
The package failed to build, but I knew right away. More than likely the concatenated version failed too and there was no message.
SIP 4.11.1 and PyQT3 3.18.1 compiled fine.
Log attached.
According to the failure error, kinputidalog.h has an undeclared class QValidator. kinputidalog.h is installed from kdelibs.
krun.cpp:977: error: 'KStringHandler' has not been declared
Does krun.cpp need an include statement to kstringhandler.h?
I notice a similar recent patch with konq_mainwindow.cc using KStringHandler. I'll guess that one will fail to compile too without the same patch?
I'm curious how these SVN changes pass muster for Debian and not Slackware. Same basic tool chain.
Do your Debian build scripts contain "make || exit 1" as a mechanism to force the make process to terminate? That is what I use to force the build process to terminate.
Darrell
How do those files work?
Do they have to recompiled against Trinity core packages or can pre-built packages be used, such as those already available in Lenny or Slackware 12.2?
Darrell
I'm seeing a whole bunch of these messages now. The [systemtray] tells me you added the messages, but the messages don't tell me much.
What area is being cleared and why do users need to know?
Darrell
Log attached.
sip, pyqt and pykde are not installed.
Last time I tried and successfully built kdebindings was svn 1175370.
Is there a way I can diff SVNs that I might help troubleshoot?
Darrell
How are messages forwarded to the list? Many times I have sent messages/replies that don't arrive in my mail box for a long, long time and other times the messages appear within seconds.
Currently I have two replies not yet in my mail box, but I already received a reply I sent after those two.
Strange or normal behavior?
Darrell
I was browsing through KControl and selected the Wireless Network module. I later saw the following messages in the xsession-errors log:
QComboBox::setCurrentItem: (protocolCombo) Index 5 out of range
QComboBox::setCurrentItem: (aclCompatibilityCombo) Index 3 out of range
QComboBox::setCurrentItem: (maxProtocolCombo) Index 5 out of range
QComboBox::setCurrentItem: (serverSigningCombo) Index 3 out of range
lo no wireless extensions.
eth0 no wireless extensions.
I don't have wireless configured in any manner. Just a wired connection.
When I selected that module, I received a warning message box four times:
[Sorry - Control Center]
Unable to detect wireless interface.
(OK button)
Four warnings match the number of tabs in the module.
I think that after the user responds to the first message box the remaining three warnings should be quashed.
Unchecking the Autodetect check box does not disable the warning messages.
I don't believe the message warnings add usability value. Possibly a better method would be to provide a message (in each tab of the module) that no connection was detected.
Further, there should be no warning messages if nothing is yet configured.
Second, the QComboBox messages should not appear in the xsession-error messages. The content of those messages indicate some kind of code error, which translates to noise and clutter in the log. Some kind of error trap is needed to quash the messages.
Third, the messages about lo and etho should be prefixed with something like [kcontrol] so users know the source of the messages. Logs are meant to help users, so let's help them. :)
I hope this helps!
Darrell
> Yes. Try reversing the kdelibs patch created by SVN
> revision r1178028
> (you can download the patchfile from the Trinity project
> website per
> usual), rebuilding and reinstalling kdelibs, then attempt
> the kdebindings
> build again.
Okay. Tried the reverse patch. Same failure.
Log attached.
In the short term I can live without python bindings because I can't get pykde to compile, but I should be able to compile kdebindings.
When compiling tqtinterface I see many warning messages like this:
tqimageformatplugin.h:43:35: warning: no newline at end of file
The messages all seem non-fatal, but adding a newline at the end of all those tqt*.h files would clean the build process. :)
Similarly, as previously mentioned a few days ago:
configure: WARNING: You are attempting to compile Trinity without the
Trinity Qt Interface installed. Please install libtqtinterface-dev
and try again!
Seems there should be some kind of simple self-check so that one message does not appear when building tqtinterface. :)
Darrell
I see these types of messages in the xsession log:
Launched ok, pid = 5906
Launched ok, pid = 5930
Launched ok, pid = 5972
Launched ok, pid = 6014
The process IDs are momentary. They are not in the ps list.
At one point I traced the messages to the KControl/Appearances/Style dialog box. I was testing the toolbar grabber handles when I first noticed the messages. Every time I selected the Apply button, one of these messages appeared in the .xsession-errors log.
However, I see the messages without accessing KControl too. The messages seem harmless, but most people do not like mystery messages in their xsession logs. Would be nice if, like we did with the startkde script, the messages provided a prefix or other clue as to exactly what launched a process.
Darrell