svn::talk 0.0.1
Filter: muesli at gmail dot com 

1614 commits found for author dfaure.
Mostly active for:   kdelibs   KOffice   kdebase   kdelibs 3.5   kdebase 3.5   kdepim   kdemultimedia

TimeProjectRevisionAuthorRepliesMessageThursday, 31. August 2006
22:12kdebase579362dfaure0Don't link to XRES_LIBRARY if it's not set
21:03kdelibs579340dfaure1787Don't use toString(); check for local urls before using path()
21:03kdelibs579340dfaure1787Don't use toString(); check for local urls before using path()
21:03kdelibs579340dfaure1787Don't use toString(); check for local urls before using path()
21:02kdelibs579339dfaure2Fix string-url confusions; use KMimeType api instead of KMimeMagic.
21:02kdelibs579339dfaure2Fix string-url confusions; use KMimeType api instead of KMimeMagic.
21:02kdelibs579339dfaure2Fix string-url confusions; use KMimeType api instead of KMimeMagic.
20:58kdelibs579337dfaure1Move url property to ReadOnlyPart where it belongs, and make it a KUrl to avoid KUrl->QString implicit conversion
20:35kdebase579327dfaure0Fix many qstring<->kurl confusions (found by making toString() and operator QString private in KUrl, which I'll commit once all modules compile with it) Made some more kurlrequesters "LocalOnly" when the code is obviously only able and willing to handle local paths.
20:31kdebase579326dfaure0FEATURE: kde-cp and kde-mv binaries.
19:30kdelibs579304dfaure0setUrl() is a bit cumbersome to use when working with paths only, so add a setPath(). Especially since giving a path to a KUrl aborts.
16:35kdebase579261dfaure0Disable kwallet (in unit test home dir)
16:28kdelibs579257dfaure0SVN_SILENT unused
15:16kdelibs579244dfaure0SVN_SILENT spaces in kDebug


TimeProjectRevisionAuthorRepliesMessageThursday, 24. August 2006
14:19kdelibs 3.5576656dfaure0Calling utf8() twice is expensive, calling length() on a QCString is expensive, so better not do either one :) CCBUG: 131242
11:30KOffice576602dfaure0One new/delete pair less. Not sure why ListItemsHelper needs a d pointer, too.


TimeProjectRevisionAuthorRepliesMessageWednesday, 23. August 2006
22:00kdelibs576352dfaure0No dbus registration if no kapp. This should fix all kio_http processes being somehow owned by kdeinit because kdeinit used kDebug (in KStandardDirs::findExe) and connected to dbus that way.
21:54kdelibs576351dfaure0SVN_SILENT no tabs
21:42kdelibs576346dfaure0With Qt-4.2 the dbusbindtoapp thing isn't needed anymore, and QCoreApp+KInstance is enough - good. This means no more kapp, so no more registration to dbus -> faster and less crap in kdbus.
16:33kdelibs576266dfaure0qWarning -> qDebug
16:32kdelibs576265dfaure0Don't use qWarning; make this test actually automated instead of just printing out stuff, and explain why the test works. Might as well delete it though.
16:12kdelibs576259dfaure0Simplify the use of dbus in kuniqueapp: instead of initializing it before forking, and then assuming the parent can use that and the child has to re-connect to dbus, let's just fork first and connect to dbus in each process afterwards. Doesn't change anything, but seems safer (and this was my first suspicion for every new kuniqueapp breakage, so let's get this out of the way).
16:09kdelibs576256dfaure0Don't use qWarning for debugging! It aborts with 'export QT_FATAL_WARNINGS=1'. Added kuniqueapptest to the automated unit tests; made it exit after 10s to avoid keeping a process lying around.


TimeProjectRevisionAuthorRepliesMessageMonday, 21. August 2006
19:07kdelibs575600dfaure0We need a QCoreApplication so that socket notifiers (as used by dbus) work (cf runtime warnings). But we also need to use it for the dbus timer, and qDBusBindToApplication still confuses me (why does kde need to call it but not qt-only apps?), so for now let's just go with KApplication. Could become KCoreApplication at some point. -> this fixes kio_http being very very slow.
19:03kdelibs575599dfaure0Cleaner way to exit dispatchLoop(); ::exit() crashes as to qdbus' qglobalstatic object with a socket notifier (which accesses the deleted event dispatcher).
18:52kdelibs575594dfaure0More qWarnings fixed
18:21kdelibs575587dfaure0Don't use qWarning for debugging. It aborts the app with export QT_FATAL_WARNINGS=1 !
18:19kdebase575586dfaure0Avoid warning from stupid QLayout when this page gets added to a stacklayout
18:06kdelibs575581dfaure0Matt should really read replies to his commits ;)
17:32kdelibs575567dfaure0This makes kde-emacs/klaralv.ev enable qt4 stuff like kDebug instead of kdDebug, uppercase includes, etc.
17:09kdelibs575556dfaure0If you make a GUI test, don't use add_test for it - this is only for unit tests!


TimeProjectRevisionAuthorRepliesMessageWednesday, 16. August 2006
19:53kdelibs573603dfaure0Fix 'kfmclient exec file:/path#anchor' as reported on k-c-d by Kevin Krammer. fdo bug 7863
19:48kdelibs 3.5573599dfaure0Fix 'kfmclient exec file:/path#anchor' as reported on k-c-d by Kevin Krammer. fdo bug 7863


TimeProjectRevisionAuthorRepliesMessageTuesday, 15. August 2006
20:20kdelibs573345dfaure0Fix "menu entries from Mathematics and Science menu are shown in Lost & Found" (#113798) due to the typical bug of using firstChild() to get the first element, which fails when there is a comment in the XML.
20:14kdelibs 3.5573344dfaure0Fix "menu entries from Mathematics and Science menu are shown in Lost & Found" due to the typical bug of using firstChild() to get the first element, which fails when there is a comment in the XML. BUG: 113798


TimeProjectRevisionAuthorRepliesMessageThursday, 27. July 2006
16:15kdelibs566942dfaure0First step: deprecate copy() for file_copy - so that dircopy can finally become copy one day.
05:40kdelibs566815dfaure0startup_id was actually missing from kdeinit_exec(); both the xml and the client code in ktoolinvocation expect it to be there. This should fix: Method "kdeinit_exec" with signature "sasass" on interface "org.kde.KLauncher" doesn't exist.


TimeProjectRevisionAuthorRepliesMessageWednesday, 26. July 2006
16:32kdebase566653dfaure0When you use a static method, don't create an instance first...


TimeProjectRevisionAuthorRepliesMessageSaturday, 22. July 2006
05:42kdelibs565017dfaure0Apply patch by Robert Hogan <robert roberthogan net> to fix untarring of the 'dante' tarball (whatever that is).
05:42kdelibs 3.5565016dfaure0Apply patch by Robert Hogan <robert roberthogan net> to fix untarring of the 'dante' tarball (whatever that is).


TimeProjectRevisionAuthorRepliesMessageTuesday, 18. July 2006
15:51kdelibs563823dfaure0Forwardport fix for 130831: remember when the last replacement text was the empty string
15:46kdelibs 3.5563822dfaure0Remember when the last replacement string was the empty string BUG: 130831


TimeProjectRevisionAuthorRepliesMessageWednesday, 12. July 2006
14:05kdelibs561564dfaure0The other half of rev 294999 by Andras Mantia (wow, two years ago): also restore m_file when saving fails. Andreas Kling noted that kate was trying to save again to the same path, when 'save as' failed once (e.g. no permissions or disk full)
13:20kdelibs 3.5561548dfaure0The other half of rev 294999 by Andras Mantia (wow, two years ago): also restore m_file when saving fails. Andreas Kling noted that kate was trying to save again to the same path, when 'save as' failed once (e.g. no permissions or disk full)


TimeProjectRevisionAuthorRepliesMessageTuesday, 11. July 2006
17:31kdelibs561005dfaure0two more uses of unofficial methods
17:15kdelibs561002dfaure0The patch re-adding isSuccess() to QDBusReply<void> is probably wrong, there's no reason for it to have a different API than QDBusReply<T>. isValid() is the correct method nowadays.
13:08kdelibs560865dfaure0Use KMimeTypeTrader correctly - should fix assert in kfiledialog found by Pino
10:30kdelibs560829dfaure2Split the big jobs (CopyJob and DeleteJob) into their own file, as discussed to Kevin (to fix kde-switch-cpp-h usage in emacs e.g.) Removed all virtual_hooks on all kio jobs since they are QObjects.
10:04kdelibs560814dfaure0Split the compiler and OS out of config.h, it's only needed by kbugreport.cpp This way when switching compilers (or kernel version) you don't get an updated config.h.
08:40kdelibs560789dfaure0Fix mimetypetrader queries not obeying the given service type, which is why "kde-open /tmp" was trying to use konq_iconview.desktop as the service to run.
08:31kdelibs560782dfaure0Same fix as in KMainWindow


TimeProjectRevisionAuthorRepliesMessageMonday, 10. July 2006
22:48kdenetwork560703dfaure0Simplify and fix compilation
22:40kdelibs560700dfaure0This includes the x11 include dirs so it fixes the build on FreeBSD, reportedly
21:15kdelibs560674dfaure0Improve error reporting when trying to Exec a desktop file without an Exec line. Next step is figuring out why konq_iconview.desktop is returned when querying for mimetype=inode/directory servicetype=Application though...
21:13kdemultimedia560672dfaure0Fix compilation
21:06kdebase560670dfaure0Enable compilation of kioclient Also compile it as "kde-open" (to be consistent with other desktop environments). In that mode the syntax becomes kde-open 'url' ['mimetype'] which is equivalent to kioclient exec 'url' ['mimetype']
19:20KOffice560609dfaure0cleanup
14:31KOffice560521dfaure0When you change stuff in kspread please remember to compile filters/kspread too CCMAIL: Robert Knight <robertknight@gmail.com>
14:21KOffice560517dfaure0Ported this one to the kspread header changes, will do the others now
13:51kdelibs560505dfaure0FEATURE: support for suspend/resume on kio jobs, and a button in the UI for it. Started this change during the kubuntu meeting in Paris, but was only able to test it today, after much qt/qdbus/kdelibs bugfixing over the past weeks :)
13:44kdebase560501dfaure0This is fun - you get an overwrite/resume dialog, and after clicking a button the app just exits...
11:41kdelibs560454dfaure0copy/paste
11:34kdelibs560452dfaure0Move KXYSelector to separate header/cpp files and move KHSSelector with it, renamed to KHueSaturationSelector.
11:15kdelibs560447dfaure0KOffice uses those classes too, not only KColorDialog. So export them. (and I renamed KValueSelector to KColorValueSelector)


TimeProjectRevisionAuthorRepliesMessageSaturday, 08. July 2006
13:50kdebase559803dfaure0SVN_SILENT trailing whitespace removal
13:49kdebase559801dfaure0Fix object path and interface name
13:47kdelibs559799dfaure0sendfile() returns ENOSYS in valgrind
13:31kdebase559794dfaure0Adjust to new object path and interface name (had to rename the interface since otherwise org::kde::Konqueror was both a namespace and a class, in generated adaptors)
13:30kdebase559792dfaure0Ported kfmclient to DBus (mostly in the bus from Trysil). The buttons in kicker don't work yet, but calling kfmclient while konqueror is already running opens a window, at least :)
11:59kdebase559775dfaure0God knows why dbus-launch exports DBUS_SESSION_BUS_ADDRESS but not DBUS_SESSION_BUS_PID - makes no sense


TimeProjectRevisionAuthorRepliesMessageFriday, 07. July 2006
09:50kdelibs559413dfaure0Sockets are sequential. QIODevice behaves all different if we don't tell it that this iodevice is sequential. This doesn't fix kio_ftp yet though, it has other issues.
09:48kdelibs559412dfaure0Remove the list of init services from ksycoca, and KService::init() and KService::allInitServices(), now that Matthias Kretz has redesigned the kcminit modules to use normal servicetypes.
08:19kdebase559390dfaure0Andre' is afraid my plane might crash, apparently, so I better commit this :) This is the tool for KIO-operations, split out of kfmclient.


TimeProjectRevisionAuthorRepliesMessageThursday, 06. July 2006
23:38kdebase559285dfaure0kfmclient is being split into two tools: kioclient (for KIO operations) and kfmclient (for talking to konqueror). I have kioclient ready here in kdebase/runtime but it uses kdelibs-trunk api (I used it in kdelibs-trunk before for testing) so it will wait until next week. kfmclient itself isn't ported yet.
17:51kdelibs559164dfaure1Fix signal/slot connection problems in the test program. The real bug I was investigating (ftp/http not working) turns out to be in QIODevice though.
12:48kdelibs559002dfaure0More updates
12:39kdebase558998dfaure0Start dbus daemon only once and only if needed - not every time kpersonalizer shows up ;)
12:26kdelibs558995dfaure0Connect to the right dbus signal
11:42kdebase558973dfaure0No such slot setText in actions anymore -> use own slot
11:28kdebase558961dfaure0Fix kdirnotify usage -> kdesktop starts! Fix porting errors - some script went a bit crazy here...
09:56kdebase558922dfaure0Don't abort on startup...
09:45kdelibs558914dfaure0Avoid warning from qt e.g. when calling kwrapper
09:35kdelibs558905dfaure0Fix kdeinit socket name in kwrapper.c, to make kwrapper work again
09:15kdebase558891dfaure0Register as org.kde.kwin (in addition to kapplication registering us as kwin_$PID) Not sure why this isn't a kuniqueapplication though (but klauncher is in the same situation anyway)
09:11kdebase558889dfaure0dcop -> dbus port; fix kill being called on an empty string.
00:01kdelibs558772dfaure0Bugs and bugs and bugs and ... sometimes bugs


TimeProjectRevisionAuthorRepliesMessageWednesday, 05. July 2006
23:55kdelibs558769dfaure0Fix interface name to match the client code... ...
23:45kdelibs558767dfaure0Export those methods to DBus. In kded modules, slots must be marked as Q_SCRIPTABLE to be exported, since kdedmodule uses ExportSlots [=> only scriptable ones]
23:23kdelibs558763dfaure1Don't require qApp, kioslaves don't have one. Improve error reporting from dbus call. Factorize memory management of the singleton into platform-independent ktoolinvocation.cpp
22:30kdelibs558742dfaure0Fix the detection of desktop files missing due to kdebase not being installed
21:52kdebase 3.5558718dfaure0icefox is right, "kfmclient exec http://www.koffice.org" shouldn't show a progress dialog
21:09kdelibs558697dfaure0Fix KArchiveTest failure spotted by coolo. There were two bugs: - behavior change in QIODevice in Qt-4.2, it does its own buffering internally now, so the derived atEnd() needs to call QIODevice::atEnd() to make sure we are done with the internal buffer too. - the mimetype detection in ktar went wrong when writing a file and that file already existed (with the wrong contents)
20:08kdelibs558674dfaure0Actually those dialogs are used in process now, not in kio_uiserver anymore.
17:59kdelibs558612dfaure0Fix compilation due to the use of None in kdialog.h
14:19kdelibs558530dfaure1Use KUrl (fix warning)
13:38kdebase558495dfaure0Move the new define (unused in this branch) to a config-X11.h file. Other X11 things should move there too, but let's do that another week.
12:20kdelibs558436dfaure0Fixes
11:18kdelibs558408dfaure0Nice one
10:48kdelibs558398dfaure0SVN_SILENT dox
10:48kdelibs558397dfaure0Another useless assert