Issue 119

Fix crash on document close properly in kdevelop (ie. without deadlocks) by first requesting kdevelop to clean itself up, then locking the smart mutex to await the finishing of any other threads already using the document, and then clearing those editor integrators of pointers to the smart interface.

Hopefully now we have a deadlock-free, crash-free document closing system :)

Added some asserts to be able to fix earlier possible bugs.
More work towards a reliable VCS interface

Improved variable-speed slewing.
Instead of a kludgy mess of nested if's to implement a triangular speed profile, I am now using much simpler code to implement a cosine-shaped profile.

There's a little bit of asymmetry in the profile (i.e., the ending speed doesn't quite get down to the level of the starting speed, so the ending is a little bit abrupt).

This is a consequence of the speed being determined for position r, and then applied over the interval (r+dr).

The solution is to make the determined speed the average of speed(r) and speed(r+dr), but I'm not sure it's worth the extra math.

Let me know if the asymmetry is bothersome, or if you have a clean solution for it...

Add a new function, GeoDataPoint::normalizeLonLat( double &lon, double &lat ) and made the map overview call it first before calculating.

The reason you would use this instead of just calling both normalize functions is that this take into accout the effects that they have on each other; eg if the north pole is crossed, the longitude should be inverted.

Also fixes some holes in the math of the previous GeoDataPoint::normalizeLat( double lat ) which allowed certain conditions to let invalid values escape.

The main practical effect of this for now is that the white dot disappearing thing now doesn't happen.

Expensive solution for a small annoying and serious
When the sidebar was hidden, then digikam closed, after restarting clicking on a tab bar button would show the sidebar at size 0 (only handle visible), which is a difficult to understand situation.

Solution:
Use a dedicated QSplitter subclass the cooperates closely with the Sidebars it contains.

Solves the bug.
When the user now moves the splitter handle so that the sidebar collapses, this is recognized and the transformed as if the sidebar was hidden by clicking on a button.

After expanding the sidebar, no size of 0 (only handle visible) will be tolerated, in case of doubt the minimum size hint is used.

The SidebarSplitter class contains also the code to save and restore state directly from a KConfig group, removing a lot of duplicate fiveliners in various classes.

Fix the 100x30 white rectangle that appears when khtmlview isn't given any HTML to parse.
E.g. in konqueror when using MMB to open something that ends up firing an application.

Approved by Germain Garand.

Fix two broken (by design) throughanalyzers by replacing the with one eventanalyzer.
WAV and AVI are both RIFF files.

This type of file can by nicely parsed in an eventanalyzer.
The empty strigita_avi.so and strigita_wav.so are installed to avoid people from reporting bugs from lingering plugins from older versions.

Make IOSlaves based on TCPSlaveBase request DNS resolution via the application.
And make the application cache results for 5 minutes.

This should avoid the DNS request storm that happens when loading webpages.
Whereas this is completly normal and has been done for years, apparently we're doing something different now that causes some cheap routers to lock up or fail to respond.

Those defective routers should be replaced, but while they aren't, we introduce a cache.

Patch by Roland Harneau.

Fix delayed emission of cached items in KDirLister so that it emits the right items (those that were known at the time of the listDir, which fixes a few race conditions), and so that we remember that this dirlister is currently listing that dir (otherwise the cache could be deleted, or the internal data structures inconsistent, leading to a crash on exit).

This fixes dolphin's folder panel not following navigation in the icon view.

Fix infinite loop which has plagued me now and then for a long time, today with the google midterm_mentor_survey.html form.

Somehow a qtextdocument can have 1 block, with 2 chars, but as text or html it would show a single space.

So atEnd was always false, move to end would stay at position 0, and MoveDown would fail.
Strange.

Fix saving of styles.
This is not the complete fix but documents can at least be saved now and we can get going with the auto tests.

The fix handles the case where StyleId is not set - this happens when the format is not based off a named style. As for reusing of automatic styles, this automatically taken care of, thanks to Qt Scribe's reuse of text formats.

Fix fucked up program logic that made Amarok crash whenever you enabled Repeat Album, and even worse, on all subsequent starts.

Note: The feature itself still isn't implemented, it will act just as if disabled.

Attempt to not crash when receiving download messages.
Not there yet, getting closer.

This commit fixes the problem with the selection in the collection browser ( and all services using the CollectionTreeView ) getting completely "stuck" when mutiple items are selected bu dragging the mouse over them

The Selection by dragging the mouse across multiple items in general is a Qt feature that is activated when multi selection or extended selection modes are specified.

This cannot be disabled if we want the ability to select more than one item at a time.

The annoyance arises because this behaviour does not mix very well with dragging.

There is a bug on the Qt bugtracker ( 197089 ) that should improve this in Qt 4.5.0.

We now have a RandomAlbumNavigator making the random album playback mode actually work! :-)

Add talkers names in history view.
Based on Aaron's patch, thanks for the patch!

The fix will be in Kopete 0.60, which will be out with KDE 4.1.

Fixed two
- Ark does not save kconfig settings at all to file
- Ark does not preserve paths by default when extracting (generally surprising users, especially after showing them the tree representation of the archive)

Work in progress; completed reflection architecture, to be fixed in details

New reflection system in place and working (great).

Better integration with kde

First appearance of context menu in New* based uml widgets.

Kickass feature of the week: highlight parsing problems when they are encountered.
Currently doesn't work on documents opened at startup (to fix)

Also I think i'll remove code duplication and get the plugin to tell the widget when new problems have been collected.

First version of BoxWidget port.
Still got to make it work from toolbar.
First commit from kde.in monsoon hackathon, yay!!

Basic Git and Mercurial support.
Implemented DVCS base class which hould be used in all DVCS plugins, it prevents plugins from duplicating code

fixed bug in git: now all system environment variables are set

fixed bug in mercurial: job<<" "<<" " or job<<"";job<<""; should be used instead of job<<"one two"

Thomas, sorry for removing your code. I was commiting to <a href="http://repo.or.cz/w/kdevelopdvcssupport.git">http://repo.or.cz/w/kdevelopdvcssupport.git</a>; because mercurial plugin had a bug (you did the same bug with job<< usage).

If you want to help with DVCS: Bazaar is not implemented yet (but it required much less job, than you did to implement mercurial).

Initial support for DVCS interfaces for Kross.

Basic Bazaar support

Support for computation of the cartesian equation of a locus (if algebraic)

Introducing variable-speed slewing.
When you move to a new position, rather than slewing at a uniform rate, the speed will gradually increase to a maximum value, then slow down as the target is approached.

Try it out and let me know what you think.
Is this appealing enough that we can just discard uniform slews, or should users be given the option?

Add semi-working colourization support

Work started on stepgame GUI

Introducing reader/writer for loading/saving KidDrawings.
Initial version of writer implemented.

First step to merging the images with similar MD5 sum during import.

added a statistics dialog that might be useful when messing with the database to check its integrity

digiKam from trunk: new dialog to show all components and libraries informations used by digiKam

a simple dataengine that provides a favicon (if available) for a given url in the form of a QImage

Improve test gui: add list of sent notifications and the ability to close currently displayed ones.

an embryo of a tabbar widget

begin of a simple button class: for now they live in the tabbar, hope to make them generic (and more pretty) buttons for plasma

Absolutely minimal proxy support with a slightly modified version of Sergey Saukh's patch from bug report 155707. This is better than nothing and a weak bugfix.

Let's see if I can do better and get a better version into 4.1...

Also there must be a better way to implement the logic around proxies than sprinkling 25 lines o'stuff into every other already-too-long function.

DBUS actions for invoking menu.

Xinerama support.

Allow switching scaling algorithm.

Native "Open With" dialog on Windows (by default).
This allows to use (and associate) given file extensions with any Windows application directly, as users expect.

- moved KRun::KRunPrivate declaration to krun_p.h
- added KRun::KRunPrivate::displayNativeOpenWithDialog() for Windows only in krun_win.cpp
- added a note to KOpenWithDialog that using KRun::displayOpenWithDialog()
is recommended

Also reported in <a href="http://intevation.de/roundup/kolab/issue2766">http://intevation.de/roundup/kolab/issue2766</a>;
(reviewed)

Beginning of a KCM to configure standard action shortcuts.

allow to resize freely
attempt to keep the text inside the applet
=> needs scrollbar as only the end lines are shown

tabbar widget that uses Tabs widget.
the test applet adds some pages, it mostly works
still needs more convenience methods and still very rough, but that is :D

Add two new modes to KDirOperator: a "tree view" mode, where you simply get a file tree, and a "detailed tree view" mode where you get a file tree and the normal extra columns (size, type, ...).

begin of a sliding animation:
seems to be totally unnoticeable, probably I will have to use a custom animation, beach..

enhanced kde termination support: terminate all running processes below the kde install prefix

First steps towards config-file support.

sliding-animates the actual tab pages
should manage removal of tabs a bit better

More work on skinning.
We have columns now.

also make sure that if the user turns off the birthdays from Contacts option, they can still see them from the birthdays resource.

separate birthdays from anniversaries as a resource so you can configure separately.

Added a 'Simple' default skin.
This wants to be backward compatible with the current KMHeaders

Added status filtering capabilities

Printing emails on Windows

The code (before patching) works on Linux (?) only by accident, I guess (probably thanks to different strategy of handling QTimers with 0 delay [2]).

Explanation: KMReaderWin::printMsg() _can_ be executed before the timer handling slotWriteNextHtmlChunk() finishes. And it is, on Windows.

Support using custom colors for Skin items.
Also use content item type tags that embed quickly testable property bits.

Added support for saving and removing calligraphy profiles.

First working kspread/ods export for kexi reports.
Currently only text is exported, but that's pretty much what you want :)

First step in preparing the paragraph tool for paragraphs which span multiple shapes.
Added the class ShapeSpecificData which stores all information about a paragraph that changes from shape to shape, for example the width of the paragraph.

This information will later be used to paint the rulers and to route input events to the rulers.

Ui for paragraph background color

Adding chalk paintop. Little cleaner code than sumi, from which it was created.

Add support to do presentation navigation from the slides list widget of the presenter view.

Finish the simplification algorithm adding additional points before calling bezierFit(), this make this function a lot more reliable, benefits of the change include:
* more accurate simplification
* avoid (virtually all cases) of bezierFit() giving totally wrong curves
* as a consequence of the above moving the guide path points is a lot more reliable, not giving weird results anymore.

Don't circle the presentation. Stop after the last page.

There is no end of presentation page yet.

Add a earned value view/chart.
Very much work in progress.

* Properly add and display abscissa
* Add all KDChart diagrams to KDChart legend
* Remove axis title shape when axis is destroyed
* Load ODF properties:
- size
- position
- title
- visibility of major/minor grid
- grid color

CSV Import Dialog
- added support for boolean (1, 0, yes, no, true, false) and floating-point data types
- fixed all dates after 2000 appearing after import wrongly as being of the 20th century i.e. 1901 or 1902 instead of 2001 or 2002; to implement this, added config value MinimumYearFor100YearSlidingWindow, by default 1930, range of years that defines where any year expressed as two digits falls.

For date window from 1930 to 2029, two-digit years between 0 and 29 fall in the 2000s, and two-digit years between 30 and 99 fall in the 1900s.

Now ladies and gentleman my last trick tonight:
->Load Cubic and Quadric Bezier curves!

implement dbus /Play GetCaps(), add details in GetMetadata().
qtscript can add menus to the settings menu

Beginnings of the bias editor UI.

Added zooming animations.

Current Track new look.
Still some alignments minor issues.

The rating widget integrated in the applet will have to wait a little bit more.
I think the change is safe but if you find anything strange then we can revert it.

Preliminary support for transferring files from any collection with a CollectionLocation to an IpodCollection.

Rating widget added to the current track applet.

The rating widget inherits from QGraphicsWidget but it's basically an adaptation from KRatingWidget from nepomuk.

It uses nepomuk's KRatingPainter. Please take a look at src/CMakeLists.txt and review if it's correct.
I've had some issues trying to add a custom star image but the default one doesn't look bad at all :P

Giant Harmony update.
It won't crash Amarok if the user enables it, it receives messages properly, it notifies the user that he needs to update his PIN, and it has 100% more comments and debug statements.

now creating QUuid for every used track and use that as url.
that way playlists still work after moving/rename of files

Now the harmony daemon stores and uses authentication data correctly, so the user doesn't have to add a new device each time they start Amarok.

Use the Album's custom context actions when triggering the context menu on the current track item.

This means no more reimplementing cover fetching actions, and gives us access to extra actions (this may be a bad thing).

eg: what place does "show under various artists" have under a menu on the current track applet?

Make the systray tooltip behave more like a real tooltip: Now we're hiding the tooltip automatically when the mouse cursor leaves the icon.

We're getting closer ;)

- This is the work I've done offline for 2 weeks.
- Added debug in Iris.
- Added JingleContentsJingleTasks, JingleSession, JingleSessionManager in Iris.
- Started JingleCallsManager in Kopete.

Allow Kopete to show accounts in status bar of contact list as it was in KDE 3.5.
Remember last opened identity status widget and show it on startup.

We need new animated icons for all protocols.

Added ChunkBar to plasma widget

Got some not altogether good animations, but it's a start

Improved task item theme: smaller borders provide more room for task item icon and text.
System tray "lasso": The relatively random system tray icons look a little better when they're sitting in their own visually defined space.

Adding a cylinder effect.
CylinderEffect extends CubeEffect and uses a shader to convert the cube into a cylinder.

This is currently very much only proof of concept and only works nearly correctly with four desktops.

Esc-key does what you would expect -> rotate to the previous selected desktop then end.
Start animation added.

Adding barebones of the KCModule

Adding real DBus support, exporting needed cmake modules

Inital version of new game bomber

Add "New" action.
Allow to open shapes from remote locations.

Added sounds to some of the games.
More sounds to come.
Please start *unmuting* your babies :) (hope I didn't break anything)

Add --fast cmdline option to turn off more preformance-intensive graphics

In Paladesign, make use of Palapeli::GameStorage, implement load/save operations, and prepare interface for the addition of a shape browser.

Make Points derive from MouseInteractor and adapt UI to allow editing of the new patternName property.
Load and save the pattern name.

Only allow to save if a pattern name has been specified. (That is important because Palapeli needs to figure out which pattern to use if there are multiple ones.

Add patch to allow playback of urls with encoded characters.
KDE/Mac rejoice.

Fix slowness in history searching for KDE 3.5

Thanks for the patch!
Sorry to take so long with it.

Looking forward to the one for KDE 4.0 so we can close this for real.

Use quicksort instead of insertion sort, results in a noticable speed improvement when loading messages in the history view widget.

* Added a tiger example C# applet.
It wasn't possible to build an executable called 'main' as mono gave an error about it not having an extension.

Maybe some sort of special cmake macro is needed for building C# plasmoids.

* Commit the results of the KDE Bindings Meeting "Let's do a smoke lib and ruby extension in 20 minutes" demo with a QtScript smoke lib and ruby extension

Kross buildtool example

Removed Mecurial plugin, because there is another one and it's hard to solve merging conflicts

Factored MultipleChoiceInput into MCInput, the base of multiple choice format input modes and MultipleChoiceMCInput, the actual multiple choice mode itself.

This allowed me to begin the article test, using MCInput as a base.

So far, the GUI works, but the AnswerValidator needs to be changed to correct correctly (heh), and fregl and I need to discuss some more about a KEduVocWordType/KEduVocArticle api change.

So, in summary: Multiple choice works like before and article test works, but you'll get everything wrong. ;)

The changes implemented in this revision will in effect move Panoramio form being dependency of marblewidget to being a independent plugin.

(jasonparser has also been moved to /src/plugin/marble/panormaio)

First pieces of work converting KidDraw to a real KDE 4 app.

- Fixing Panoramio support.

- Disabling the Panoramio support as it currently only shows two pictures and as it doesn't update the view.
Otherwise it's a nice start already.

Removing custom paint engine, not going to use it.

Update plasmoid with a couple things:
* Various stuff, including taking out some stuff which doesn't work
* Remove some debug spam
* Add code to make sure that the Marble compass, scale bar, etc are hidden

Still todo:
Fix weird painter bug

This commit removes all of the Plasma stuff.
After the WoC work in Plasma I never found the time to make the two plasmoids work again.

Better luck for 4.2, I hope.

Removing apogee & fli libraries.
They are now optional components of KStars.

I've managed to package apogee, fli, and nova libraries using SUSE Build Service and they are now available for multiple distributions via one-click install.

Third step of refactoring the Import Dialog.
Now I've introduced a Settings object that brings data from the dialog to the handler which does the actual import.

Almost there, and ready to introduce the new feature that started all this...

This completes the ImportDialog refactoring.

On UNIX, make use of a pty process instead of a normal process, as unrar opens pty connections for getting passwords.

In other cases, when getting password requests via stderr, just kill the process and return with a failure code: the proper solution would be handling the password request, but this is a non-trivial task that require some more work.

Thus, for now, just do not allow password-protected RARs to be opened.

new home designer ( Ogre + Blender import)

Add pause, read to end, forward and back toolbar icons.
Start to implement read to end, forward, and back.
Needs immediate curve updates.

one of those little things that bugged me: we have this beautiful list of themes in the settings dialog, but we use the name *on disk* rather than the name in the .desktop file and they were unsorted.

ew!

much nicer now!
heck, you can even tell that "default" is actually "Oxygen"! =)

add a kconf_update to add the Alt+F1 keyboard shortcut to the user's app launcher (well, kickoff or simple launcher anyways)

no wonder people on suse were missing the toolbox; it's pretty much invisible with aya.

still need something more visible, i think, but this is better (as in: moderately visible) and we can visit it again in 4.2

spamming my log files with "Protocol not supported" is not acceptable.

this needs to have a more selective mechanism for reporting or simply not report at all.

* prevent removing active containments (brute force mechanism: just ignore the request totally)
* select the correct containment for removal when zoomed out (will have isses on multi-screen; see FIXME)
* move containment destruction logic out of Corona; it really doesn't belong there
* remove the hack to not set focus on a containment when clicked as that doesn't seem to reliably trigger it for me here and it doesn't address the issue of a new containment still messing things up, so we really ought to have a generic solution instead of a slightly ugly hack

complicate the loadDefaultLayout stuff a bit by doing it a bit more manually; however this results in no eye-jarring visible applet resizing and jigging about when starting from a new set up.

first impressions are everything right?

Reenable the SSL dialog when clicking on the (recently fixed/reintroduced) shield icon in the URL bar.

Reusing the string "SSL" for the name due to string freeze - is there a better way to deal with the freeze?

Moved Nepomuk filewatch service into kdebase as discussed on kde-core-devel

that brown paper bag is for me and for the next 24 hours.

Enable the mouse tracking plugin by default.

disable procmon on archs where it is still unsupported.
The code surrounding it is very architecture specific and the current tests isn't precise enough to find the right archs when the archs is non amd64 non i386 non ppc.

Patch by Modestas Vainius, approved by johnflux.

Don't use big pushbuttons for arrows.
Use svg arrows instead (borrowed from rssnow applet) - perhaps they'll need some improvement to be more visible

Port KPixmap cache to QFile's memory mapping features, which work on Windows too

Moved charselect applet to kdereview

Reduce spin box step size for standby/suspend/off setting from 10 minutes to 1 minute.

This should make it easier to change the values for the average user.
Some users might be frustrated by 10 minute steps and think they can't use smaller steps otherwise.
It also saves the user from using the keyboard for changing the settings.

5 minutes might be a more practical value, but it is less intuitive for the average user.

Entering high values is probably more rare and can be done using the keyboard.
But the limit is currently at 120 minutes anyway.

lets risk it and call Dolphin version 1.1

don't rubber band from the title area; allows it to be dragged by the title as a bonus, something i kept trying without thinking

merge the vi input mode branch back with trunk again

No configurable shortcut for
About KDE
About <application>
Open Recent

Now that we have a "Default Keyboard Shortcuts" Module the name "Keyboard Shortcuts" is ambiguous.
So rename it to "Global Keyboard Shortcuts".

Please feel free to propose better names.

initial commit of Kommodity/GIO (Qt/C++ bindings for GIO/GVFS)

Major cleanup to make the code a bit tidier and to match the decibel coding style, since that is where this applet will most likely end up when it moves out of playground.

Previously there was no consistent code style, just mess...

initial commit of the server hotlink plasmoid

- change the layout to reflect the results of the diploma thesis implementation, just so we don't lose SVN history
- remove old test files

Don't ask "open (embedded) or save" if we have no part to embed anyway.
This fixes the double-question (open or save + open with ark or save) when clicking on a tar.gz in <a href="http://www.exiv2.org/archive.html">http://www.exiv2.org/archive.html</a>;

I keep wondering about merging those two into a single "open embedded / open in app / save" dialog...

Import the results of the diploma thesis "Fusion von Web Services mit lokalen Diensten" ("Fusion of Web Services with local services") by Thomas Friedrich.

With these significants enhancements, the D-Bus Web Service Proxy is now usable for the vast majority of services. For more info, visit <a href="http://techbase.kde.org/Projects/D-Bus-WS.">http://techbase.kde.org/Projects/D-Bus-WS.</a>;

Note that the same files are still inherently GNOME-specific, volunteers for KDE-related D-Bus services are welcome.

Apply patch by Johannes Simon to improve usability of search bar:
"When a query is started, I think it's better not to delete the contents of the search box. Instead, the contents should be selected when a different search engine is chosen, so that the user can either start overwriting the text with a new search text right away, or hit enter and do the same search again with a different search engine. This btw is exactly the way it's done by Firefox' search bar.

Currently, I find it really annoying having to re-enter the text over and over again when I want to redo the search (either with the same search engine, or another)."

Get rid of kleopatra and libkleo, will be copied over from trunk in the next commit.
This is easier than solving the merge mess, and we don't need that accurate merge info in the trunk -> soc direction anyway.

However, I'll mark every kleo-related commit as merged.

a smile-face icon.
sorta cheesy, but the old cookies icon is gone and something is probably better than nothing.

Apply the skin columns when needed.
Distribute the available space by using the column size hints and some heuristics.

Great mother of pearl it feels good to hack again!

- Fixed lack of error checking after we try to open KDE wallet in keyring conduit. Running KPilot from trunk in a GNOME environment (don't ask) fails at opening our wallet, but SIGSEGV'ing is never a nice thing to do.
- Read up on new Qt4 threading model and broke our previous worker thread into a pure thread class and a pure worker class.

This still needs LOTS of cleanup and refactoring and love.
I'll try to get to more of this this week.

QTimers now belong to pure worker class and UI looks perfectly responsive, altho I'll wait for Bertjan to prove me wrong.
- Ran several syncs in succession using "usb:" device, which is notoriously picky.
Looks good!

bring the status bar style preview back

* Completed port to pion-network-library
* Removed unused files

Testing basic design. I paint "2D gaussian distribution" brush on canvas.

Some tests with sumi-e, ink is lost really fast, bristles are randomly omitted.

explicitly use bash as shell as we use quite a few non-standard commands

will be part of upcoming kile 2.0.2

* local copy of shttpd no longer required

Refactor to make this work propertly :)

The KoChartModel now is a QObject as qobject_cast is the only thing that works in this case. The problem is that KoChartModel is not a class in a shared-library, but a class compiled into each using-application.

See <a href="http://techbase.kde.org/Development/Tutorials/Common_Programming_Mistakes#dynamic_cast">http://techbase.kde.org/Development/Tutorials/Common_Programming_Mistakes#dynamic_cast</a>;

The problem we had is that KSpread wanted to use QAbstractTableModel but kchart wants to use the extra implementation of QStandardItemModel. So the change has to be from 'is a' to 'has a' model.
The effective changes are not that big; kspread now has a new class (oddly named) that is the model in Binding_p.h and kchart simply has a QStandardItemModel member.

I suggest someone come up with some better naming now since KoChartModel for a class that owns a model, but is not a model might be confusing ;)

tags koffice 1.9.95 Alpha 9

tag katelier

Plugins
Calendar

Start to port the calendar plugin from a KPart to a cell tool plugin.

Ok, I think I'll not understand the mess with KoListStyle soon, and I hate having code waiting on my harddrive for more than a week.

I store the outline level in KoTextBlockData because it is *not* a style property.

So I agree that it could be cool, easier to have the outline level in the style, but it is not the right place if we want a strict implementation.

Setting the outline level in a block attribute would have bad results : the QTextBlockFormat could be different for blocks even if they share the same style, and that would be a mess to merge back the QTextBlockFormats together.

Because it is perfectly legal to have two headings with the same style but a different outline level. For instance, the user create a "heading" style, apply this style for every heading he has, but decides to change the outline for each heading. It is allowed (but not supported by OpenOffice.org).

Also supporting the default outline level is gonna be quite hard, because we need a difference in KoParagraphStyle::applyStyle : if we apply when loading the document, the default outline level doesn't matter (see the spec quote above), but if we apply following a request from the user, we must set the outline level if it's not already set.

I hope this explanation is clear enough.

Ideally, I would have used a KoListStyle, but I fear we will run into troubles when saving. (Ok, currently, we don't support saving lists at all, that's easier)

So, I'll continue hitting the wall with my head to understand how we are supposed to use KoListStyle with headings and still support the spec, but meanwhile I commit to show that I'm working on this thing and to prevent any code loss (I had an hardware problem yesterday, I only hope I won't have other problems in the next days).

Redesign the shortcut system for activating tools.
We used QActions which have the disadvantage that they go in front of everybody, which meant that a krita shortcut without things like Ctrl would make it impossible for users to type that key in other places.
This new solution expects a tool to property 'accept' used keys and all the keys that are unaccepted are now scanned for a match with the tools keyboard shortcuts.
If there is a match, we activate the tool.

This limits the keys to activate a tool from being application-wide to only working when the canvas has focus.

I'm convinced that this is the only sensible way we can actually support one-key shortcuts, though.

add initial work on the visual changelog for Alpha9

Start porting Now Playing script to MPRIS.

Remove the "Stop" action menu, which provided the following two options: "Now" and "After Current Track".

Reasoning for removal:
* It added an extra sub-menu in the systray menu. When you pressed Stop, another menu cames up, which was a hassle.
* It added an extra arrow in the main toolbar, which looked quite ugly and confusing.
* Seems that the "Stop After Current Track" feature was not very popular.

If anyone is against the removal, please speak up now.

Revert <a href="http://websvn.kde.org/?rev=815221">r815221</a>;, which tried to use the native plasma WebContent thing for the Wikipedia applet, but had horrible side effects, like images not showing and borked scrolling.

figured out why Fisheye said Amarok suddenly jumped to 1-2 million LOC:
this 2.6 megabyte file.

Peter!! :)

Ported backend support for copying tracks to Ipod, to Collection-based stuff.

Make the Current Track applet use an internal copy of the KRatingPainter from Nepomuk.

Now we no longer have to depend on Nepomuk.

Plastering my name everywhere.
This is my legacy...

now I can croak and know i've left something behind for the children. *sniff*

fix typo: NEPOMUTACK -> NEPOMUKTRACK.

How many funny variations of NEPOMUK are there? :)

rework some of QueryMakers api.

instead of calling startXQuery (which never made much sense to me as a construct, as it doesn't actually begin a query) create a new method, setQueryType that takes an enum of various types.

In the future we may want to transform this into a bitfield and allow for constructs like setQueryType( Artist | Album ) but thats a future question.

This will allow for the removal of the enums in most private classes that do this same thing at some point in the future

Remove some KDE 3 and Qt 3 compat code.
Can you believe that I had a dictionary mapping QStrings to pointers-to-int?

Me neither...

Put Ramblurr's DebugBlockify script in SVN.
I figured it could be useful to others too.

Are you OK with that, Casey?

Remove the dev-4.2 branch.

I have all the break-the-contactlist changes saved and will reapply them. I need to recreate this branch though in order for svnmerge to work properly (apparently).

Add dev-4.2 which is a development/integration branch for things that are already done and need to be merged for 4.2 or are still incubating and might be done if I have enough time. :)

Copying bonjour protocol from extragear to trunk (code will be in extragear until KDE 4.2)
Still to edit CmakeLists.txt in protocols dir to make this compile

ported meanwhile support to KDE4.
Few TODOs left (status message not handled, better error reporting).

Still not enabled in default build, need to test it first.

Added Bonjour To kopete (again)

Close cube with release of right mouse button. (I do not understand why catching a double click event does not work...)

last minute icon

We only need window quads in cylinder - not in cube.<brCylinder vertex shader is working correctly for every number of desktops.

There is still a problem with odd number of desktops.
This has to be addressed in the effect.

Inital commit, no grouping support yet

Opening Powerdevil Branch. Powerdevil will be a KDED-based Daemon for laptop powersaving functions.
It will consist of a KDED module (the main core), a KCModule to configure it, and an applet to monitor the various states.

It will have some backported code from kde4powersave

If the sound feature is not ready for release, disable it completely, including removal of the sound item from the Settings menu.

update to *almost* all ksirk icons and default graphics.
Please comment.

new icon for Kubrick game.
comments are welcome.

Enable the sound feature, but leave the "Play Sounds" setting turned off by default.

Default to a 4-way gradient for the background if loading the SVG file has failed.

Import of Paladesign, an editor for patterns.
It is rather useless by now because of missing load/save functionality and because an implementation of the generated patterns in Palapeli is missing.

bring fibs chat to so-so from terrible

Start of a refactoring.
The code has at some points turned into quite a mess.

Some class interfaces need a redesign and cleanup, esp. to speed up compilation and prepare for the introduction of the pattern plugin architecture.

So here we go:
Refactoring, part 1: Turn Manager into a singleton, access Manager through ppMgr() from all other classes.

Draft for the site utils.kde.org

Should be usable as-is already to put it online.

I concentrated on projects/okteta to give an example what data and how should be presented IMHO.

For the other projects I added just some basic data as available from the KAboutData and the abstract in the manual.

Everybody is encouraged to improve the pages {index,faq,features}.php for the other projects. While this site draft stays in playground, everybody with general commit rights can do, so please do not hesitate :)

Screenshots are still missing, can be added later, should go to the feature page.

Please see utils/README how to setup this site

Remove the "File a new bug" link from the bug list

It would appear when doing a search or looking at the bugs for a particular opponent. We don't want it. We got a wizard.

Branching of KDE 4.1

imported a reworked version of BTL (Benchmark for Templated Libraries).
the modifications to initial code follow:
* changed build system from plain makefiles to cmake
* added eigen2 (4 versions: vec/novec and fixed/dynamic), GMM++, MTL4 interfaces
* added "transposed matrix * vector" product action
* updated blitz interface to use condensed products instead of hand coded loops
* removed some deprecated interfaces
* changed default storage order to column major for all libraries
* new generic bench timer strategy which is supposed to be more accurate
* various code clean-up

add benchmark for slice vectorization... expected it to be little or zero benefit... turns out to be 20x speedup.

Something is wrong.

raah, results were corrupted by overflow.
Now slice vectorization is about a +25% speedup which is still nice as i expected zero or even negative benefit.

KDE 4.1 RC1