Issue 202

Store full path to kdevelop executable in the x11 session restore command.
If you have two or more kdevelop's installed (for example, self-compiled in home dir
and distribution-provided in /usr), session restoration will launch the
version you actually used. Before it would launch only the one in /usr.

Allow kdesrc-build to remove existing source dirs.

That is, if the user permits it using --delete-my-patches, and only for
git clones at this point.

This should hopefully resolve bug 282367 (also frequently reported on
IRC ;)

fix editing of factorisation tasks

Now it is possible again to just use the keyboard to the factorisation
exercises.

Fix was contributed by Luis Angel Fernandez Fernandez. Many thanks to him for
fixing this long standing bug!

BUG: editable matrixes could not load their name info. Now they can.
BUG: on reload from a session file, plot ranges were not respected if zoomsr
were tied when saved. Now they are

FileLoader: Recenter on file bbox when single user file

Plasmoid: do not show Terrain data as it irritates some

Override ck check if the last active user requested to shutdown the session.


FIXED-IN: 4.7.2

Fixed bug for selecting files with --select parameter.

Files selected via --select parameter did not get current item focus.
And if files were deleted while being current item, updating the view
would select the next item instead of the first item in the list.

Make sure we only run one instance of nepomukindexer at the time.

This is more of a workaround/fail-safe in case some of the asnyc code
is buggy and results in several instances of nepomukindexer.

Refine the 'Change profile' action.

- Rename it into 'Switch Profile' to avoid potential confusion.
- Also put it under the 'Settings' menu.

TODO: rename related methods which use 'change' in their names.


FIXED-IN:4.8
REVIEW: 102686

fix memory leak in solid:

The initial visibility of menubar should be determined by profile option.


FIXED-IN: 4.8
REVIEW: 102502

Fix the problem of possibly losing the last line after resizing window.

This is a simple one-line patch, so it can be reverted easily if
unfortunately needed.


FIXED-IN:4.8
REVIEW:102177

Never assume the query returns a result.

This fixes a crash if there is some query error.

Repaint the whole window in the taskbar thumbnail

this really kills my windowstrip artifacts

Delete the service before quitting the application.

This is required since the service shutdown sometimes requires a fully
funcational QCoreApplication. Not having that leads to crashes.

Use a lock file to indicate the screen is locked

The lock file is touched when the screen gets locked. If the
file cannot be created the screen won't lock. If there is the
lock file when KWin starts up, the screen locker will be activated.
This ensures that KWin will safely lock the screen even after a
crash.

The lock file is set in the cache standard dir. This should be
moved to the run standard dir after it has been implemented.
The lock file is owned by the user and not writable by group or
others. This means only the user can remove the lock file and by
that end the screen locker when KWin crashes/restarts. As it
requires a local access it should be save enough. It might be
possible to make it owned by root and use something like polkit
to provide the security, but that is something someone more
competent in that area has to decide. The previous implementation
allowed to end the screen locker by killing a process run by the
current user, so it is as secure as the previous implementation.

Apply an overlay notifying the user the Power Management service is not running.

Whenever the daemon is not running, we don't want to let the user edit anything, as the config itself
relies on the daemon.


FIXED-IN: 4.7.2

Don't show the default profile in list when no others are listed.

When user selects only 1 profile to be listed in menu, it makes the
'New Tab->' list unnecessary. No profile will listed if the only
selected profile is the default one.

Patch by Francesco Cecconi


REVIEW: 102670
FIXED-IN: 4.8

kwin: bugfix for taskbarthumbnail and slide effect interplay

This patch avoids that the thumbnails slide when the desktop is changed.

REVIEW: 102675

fix close button side for present windows effect


pint desktop as background when including desktop in switcher


zoom windows as hover/selection indicaton (1/8 of the screen or 105%)



no closer on "show desktop" desktop
show closer immediately but have it disabled for a short time to allow the user realize it

REVIEW: 101318

Conflicts:

kwin/effects.h

only try to change the name once

Set a program name to show a nicer name in the kwallet dialog, not just "Akonadi Resource".




Conflicts:

akonadi/agentbase.cpp

Ignore items with empty remote ids here.

Items with empty remote id are those not yet stored by the resource,
ie. we still have pending change replays for those. So, if we do a sync
during that period (which could happen since the introduction of the
file system watcher), we would remove a random one of those (only one
since the map would only contain one without a remote id). This mostly
affects the POP3 usage pattern.

Fix Bug 227582 - kmail ignores disabled auto-line-break-settings (in menu-checkbox and global config)

FIXED-IN: 4.7.2

Fix saving that i had broken in previous commit

also fixes together with previous commit

Fix Header and Footer are missing in the attached document
Another interesting undocumented finding; If a paragraph references an non-existing
page-style name then OO.org falls back to using the display-name as identifier. Now
we do the same.

rename KoShapeControllerBase to KoShapeBasedDocumentBase

KoShapeControllerBase is _not_ (aha! got you!) the base class for
KoShapeController, instead it's an interface that document classes that
own shapes can implement to allow adding/removing shapes to their
internal shape data structure.

In other words, it's a base class for documents whose content is based
on shapes, in other words, a KoShapeBasedDocumentBase class...

Until 2006, it was called KoShapeControllerInterface, which made
sort of sense.

As discussed on irc:

15:29:42 < boud> I'm getting completely confused between KoShapeController and KoShapeControllerBase...
16:04:43 < sebsauer> boud: a good indicator that they are bad designed
16:10:03 < boud> yes, I totally agree
16:12:10 < svuorela> they need a AbstractKoShapeControllerBaseInterface to derive from
16:13:42 < boud> with one implemenation: SimpeKoShapeControllerBaseInterfaceImpl
16:14:14 < boud> I was browsing the git log, and it turns out that last summer I was hurting because of this design as well
16:15:53 < boud> actually, KoShapeControllerBase used to be called KoShapeControllerInterface, but that was renamed in 2006
16:17:19 < boud> jaham called it "Interface" -- and that still seems more accurate, since most methods are pure abstract. Thomas Zander renamed it.
16:24:49 < boud> I really wish I could get the shapecontroller right out of the canvasbase
16:27:18 < boemann> boud: to be far the renaming was idscussed on the ml and base is if it also implements base behaviour
16:27:22 < boemann> be fair
16:27:43 < sebsauer> lol re AbstractKoShapeControllerBaseInterface ... maybe turn it to a AbstractKoShapeControllerBaseInterfaceFactory ?
16:28:11 < sebsauer> or even a AbstractKoShapeControllerBaseInterfaceFactorySingleton
16:28:17 < boemann> sebsauer: no a factory cant return interfaces :)
16:28:27 < sebsauer> boemann: in C++ it can :-)
16:28:38 < sebsauer> interfaces without pure virtual
16:28:40 < sebsauer> 's
16:28:51 < boemann> ok it can but not accoring to our defs
16:29:22 < sebsauer> but yes, it's not nice and also misses a strategy-pattern
16:30:01 < boemann> AbstractKoShapeControllerStrategyBaseInterfaceFactorySingleton
16:30:16 < sebsauer> sounds like a good name
16:30:30 < boemann> ok boud go ahead implementing our new design ;)
16:30:34 < sebsauer> :)
16:32:05 < svuorela> it's especially a good name for people who are paid by character ...
16:32:31 < boemann> :)
16:32:53 < boud> boemann: but this class doesn't implement the base behaviour for KoShapeController...
16:33:10 < boud> boemann: it's kind of an interface a document class implements so KoShapeController can add/remove shapes to the document
16:33:28 < boemann> oh
16:33:33 < boemann> nasty
16:33:43 < boud> the ***base renamings were last year, this was already in 2006
16:33:55 < boemann> ah
16:34:15 < boemann> i'd say we need to rename it
16:34:25 < boud> and every time, every time I look at these classes I am just as confused as you were just now -- surely KoShapeController must implement KoShapeControllerBase?
16:34:55 < boud> maybe KoShapeControllerDelegate?
16:35:21 < boemann> yeah
16:35:41 < boemann> if not a completely different name
16:36:25 < boud> it also creates a KoResourceManager and sets that on the shape factories for some reason... need to dig in again
16:36:34 < sebsauer> Ko42
16:37:20 < boud> lol
16:37:59 < svuorela> that's a great name. and then it can be subclassed to Ko43 when more virtuals are needed
16:38:26 < boud> how about KoShapeBasedDocumentBase...
16:38:53 < boemann> fine with me
16:39:05 < sebsauer> :-)
16:39:23 < boud> ok, will do
16:39:32 < svuorela> KoBasedShapeBasedDocumentBasedBase
16:39:32 < boud> I can't think of anything better
16:40:08 < boud> done..

Fixed input mask for date, time and date/time data fields in table views

todo: finish this is forms

DIGEST: ^^

Implement a high-level paste method in KoTextEditor

This pastes the selection from one KoTextEditor in another.

Bump the Amazon API schema version to the latest one (2011-08-01)

Also add in some placeholders for upcoming addition of ES, IT, and CN
locales in Amazon search

rename KoShapeControllerBase to KoShapeBasedDocumentBase

KoShapeControllerBase is _not_ (aha! got you!) the base class for
KoShapeController, instead it's an interface that document classes that
own shapes can implement to allow adding/removing shapes to their
internal shape data structure.

In other words, it's a base class for documents whose content is based
on shapes, in other words, a KoShapeBasedDocumentBase class...

Until 2006, it was called KoShapeControllerInterface, which made
sort of sense.

As discussed on irc:

15:29:42 < boud> I'm getting completely confused between KoShapeController and KoShapeControllerBase...
16:04:43 < sebsauer> boud: a good indicator that they are bad designed
16:10:03 < boud> yes, I totally agree
16:12:10 < svuorela> they need a AbstractKoShapeControllerBaseInterface to derive from
16:13:42 < boud> with one implemenation: SimpeKoShapeControllerBaseInterfaceImpl
16:14:14 < boud> I was browsing the git log, and it turns out that last summer I was hurting because of this design as well
16:15:53 < boud> actually, KoShapeControllerBase used to be called KoShapeControllerInterface, but that was renamed in 2006
16:17:19 < boud> jaham called it "Interface" -- and that still seems more accurate, since most methods are pure abstract. Thomas Zander renamed it.
16:24:49 < boud> I really wish I could get the shapecontroller right out of the canvasbase
16:27:18 < boemann> boud: to be far the renaming was idscussed on the ml and base is if it also implements base behaviour
16:27:22 < boemann> be fair
16:27:43 < sebsauer> lol re AbstractKoShapeControllerBaseInterface ... maybe turn it to a AbstractKoShapeControllerBaseInterfaceFactory ?
16:28:11 < sebsauer> or even a AbstractKoShapeControllerBaseInterfaceFactorySingleton
16:28:17 < boemann> sebsauer: no a factory cant return interfaces :)
16:28:27 < sebsauer> boemann: in C++ it can :-)
16:28:38 < sebsauer> interfaces without pure virtual
16:28:40 < sebsauer> 's
16:28:51 < boemann> ok it can but not accoring to our defs
16:29:22 < sebsauer> but yes, it's not nice and also misses a strategy-pattern
16:30:01 < boemann> AbstractKoShapeControllerStrategyBaseInterfaceFactorySingleton
16:30:16 < sebsauer> sounds like a good name
16:30:30 < boemann> ok boud go ahead implementing our new design ;)
16:30:34 < sebsauer> :)
16:32:05 < svuorela> it's especially a good name for people who are paid by character ...
16:32:31 < boemann> :)
16:32:53 < boud> boemann: but this class doesn't implement the base behaviour for KoShapeController...
16:33:10 < boud> boemann: it's kind of an interface a document class implements so KoShapeController can add/remove shapes to the document
16:33:28 < boemann> oh
16:33:33 < boemann> nasty
16:33:43 < boud> the ***base renamings were last year, this was already in 2006
16:33:55 < boemann> ah
16:34:15 < boemann> i'd say we need to rename it
16:34:25 < boud> and every time, every time I look at these classes I am just as confused as you were just now -- surely KoShapeController must implement KoShapeControllerBase?
16:34:55 < boud> maybe KoShapeControllerDelegate?
16:35:21 < boemann> yeah
16:35:41 < boemann> if not a completely different name
16:36:25 < boud> it also creates a KoResourceManager and sets that on the shape factories for some reason... need to dig in again
16:36:34 < sebsauer> Ko42
16:37:20 < boud> lol
16:37:59 < svuorela> that's a great name. and then it can be subclassed to Ko43 when more virtuals are needed
16:38:26 < boud> how about KoShapeBasedDocumentBase...
16:38:53 < boemann> fine with me
16:39:05 < sebsauer> :-)
16:39:23 < boud> ok, will do
16:39:32 < svuorela> KoBasedShapeBasedDocumentBasedBase
16:39:32 < boud> I can't think of anything better
16:40:08 < boud> done..

Implement a high-level paste method in KoTextEditor

This pastes the selection from one KoTextEditor in another.

Fix bug causing UPnP not to work properly if there is a question mark in the controlurl

Make mounting (and unmounting) work under FreeBSD.

Enable hover effects on system tray icon.


FIXED-IN: nm09

Prevents two crashes because of dangling pointers when NM stops.

Don't crash if the rekonqui.rc file can't be found

REVIEW: 102673

Fix crash in dht::TaskManager and reformat source of several dht classes

Allow loading of icon themes that contain blank space in names.

Only save the notes if the contents are not empty

This fixes a very severe bug which would cause the notes to be
saved as empty notes. I've lost all my notes from the last 10 days
because of this stupid bug.

The QTextDocument would emit the modified signal, on calling reset(),
which would result in the note being re-saved, but since the contents
are now empty, it would be saved as an emtpy note.

Add Patch to avoid POP Mail loss

Add --delete-my-patches option.

This doesn't so much actively search for things to delete. Instead it's
a required permission from the user to remove source directories (which
is similar to the remove-after-install option).

Right now this will only be used when trying to clone a git module and
the destination source directory already exists somehow, which has
happened for a lot of 1.14 early adopters due to old kde-baseapps
checkouts having a kate or konsole directory that conflict with where
kdesrc-build wants to put the current kate/konsole directories.

fix to enable a new game with the same players of the last game without having to add the players again

South Sudan is here

Add action to jump to last tab.

Allows user to select a shortcut to jump to the far right tab. There
is no default shortcut.

Patch by Francesco Cecconi
REVIEW: 102687

FIXED-IN: 4.8

support 'sleep', 'hibernate' commands

Adding an explicit ScreenLocker class

This class handles just the screen locking, but not the screen
saving. This are two different concepts which have to be separated.

The ScreenLocker takes care of locking the screen by e.g. the
shortcut and invokes for example the legacy screen saver engine to
show the saver and "lock" the screen.

The ScreenSaverEngine is hardly changed. It just forwards the lock
request to the ScreenLocker to let the locker decide how to lock.
The Saving functionality itself is not changed at all including the
D-Bus interface which has to stay in the saver as it is modeled
around screen saving and not screen locking. We probably need a new
fdo D-Bus interface for that.

In the next step it is possible to move the locking into an effect
to no longer require an X screen saver to perform the locking which
would immediately remove all the glitches we see with the screen
savers.

Also in a next step we need to ensure that the screen stays locked
even if KWin crashes (e.g. through a lock file).

Architecture to provide screen locking in an Effect

When the screen gets locked Effects are asked whether they
provide Screen Locking functionality. Exactly one effect is
allowed to provide the lock screen and to unlock the screen
again.

Next step: write an effect.

Implement some missing sorting options

Changing the sort order and enabling/disabling the
"Sort Folders First" option works now.

Autoselect the device if the address match

Before in BlueDevil 1.0 the discovering page could be skipped but
in 1.2 it does a lot more, so instead of skipping it lets
autoselect the desired device.

Apply an overlay notifying the user the Power Management service is not running.

Whenever the daemon is not running, we don't want to let the user edit anything, as the config itself
relies on the daemon.

Added option to show keyboard accelerators when Alt key is pressed (not implemented yet)

Implement 'Sort By Size'

It only works for files so far. The sorting of folders
by the number of items is more tricky to get right because
this number is retrieved asynchronously by
KFileItemModelRolesUpdater.

ScreenLocker needs to be a fullscreen effect

Even if it's fun to start PresentWindows from the screen locker it
is not exactly what you want.

Changed proxy handling in HTTP ioslave to support multiple proxy server, i.e.
PAC script files and fix several other related issues:

* Added a new function to TCPSlaveBase that attempts to connect to the a
remote server without automatically sending error notitification to the client.

* Moved the proxy related code from 'resetSessionSettings' to 'httpOpenConnection'.
Proxy information will now only be set from 'setHost' and only be reset when
'reparseConfiguration' is invoked.

* Added a new variable to the HTTPRequest to store multiple proxy URLs.

* Modified 'httpShouldCloseConnection' to account for multiple proxy addresses.

* Modified 'sendQuery' to connect to the remote server before formatting the HTTP
headers so that the headers properly reflect the correct proxy settings.

* Removed support for 'PersistentProxyConnection' option since it serves no
other purpose than to cause unnecessary failures and bugs when using proxy
servers. To see these problems, simply disable the 'Use persistent proxy conneciton'
option in the proxy configuration dialog and use a proxy server, e.g. privoxy.

* Do not reset m_request.ikeepAlive flag in httpCloseConnection. Otherwise, it
can inadvertently set the flag to false in ::sendQuery where httpCloseConnection
might get called under certain circumstances. In general this flag should
originally be set in one location, resetSessionConfiguration, and changed
only to accomodate what the server based on its response.

* If the server sends back a "keep-alive" header, but reports itself as a
HTTP/1.0 server, make sure m_request.isKeepAlive is set to true.

REVIEW: 102696

* Added code that resolves a request url's hostname before attempting to match
it against the no proxy for list so long as the "ResolveHostNamesBeforeProxyCheck"
option is set.

* Allow "DIRECT" as a special keyword in the list of proxy server addresses
returned in slaveProtocol(const QString& protocol, QStringList& proxy).

* Modified KProtocolManager::proxyFor to properly handle the changes in the new
proxy management dialog (KDE 4.8) where the proxy server port, in the
manual proxy configuration mode, will be saved separated from the address with
a white space.

* Moved the code that accounts for SOCKS proxy from KProtocolManager::proxyFor
to KProtocolManager::proxyForUrl where it belongs. The current implementation
only works correctly under one circumstance while breaking the previous behavior
of the function.

* Fixed KProtocoManager::proxiesForUrl so that it accounts for the proxy
exception list.

* Update API documentation.

REVIEW: 102691

* Added code that resolves a request url's hostname before attempting to match
it against the no proxy for list so long as the "ResolveHostNamesBeforeProxyCheck"
option is set.

* Allow "DIRECT" as a special keyword in the list of proxy server addresses
returned in slaveProtocol(const QString& protocol, QStringList& proxy).

* Modified KProtocolManager::proxyFor to properly handle the changes in the new
proxy management dialog (KDE 4.8) where the proxy server port, in the
manual proxy configuration mode, will be saved separated from the address with
a white space.

* Moved the code that accounts for SOCKS proxy from KProtocolManager::proxyFor
to KProtocolManager::proxyForUrl where it belongs. The current implementation
only works correctly under one circumstance while breaking the previous behavior
of the function.

* Fixed KProtocoManager::proxiesForUrl so that it accounts for the proxy
exception list.

* Update API documentation.

REVIEW: 102691

Architecture to provide screen locking in an Effect

When the screen gets locked Effects are asked whether they
provide Screen Locking functionality. Exactly one effect is
allowed to provide the lock screen and to unlock the screen
again.

Next step: write an effect.

Conflicts:

kwin/effects.cpp
kwin/effects.h

Implement blacklist in config to not show apps

This is in preparations for Plasma Active One, where we do not want
to show everything, its dog, and the fleas on the dog's tail.

A comprehensive list of blacklisted apps is going into /config/.

Same is there for apps as well.

The config file looks like this:

[blacklist]
apps=minesweeper,dolphing,gimp,
plasmoids=org.kde.active.nonsense,org.kde.active.nonnonsense,org.kde.inactive.noplasmoid,

and so forth.

First working version of the emailfeeder plugin (replacing the old feederagent)

Still missing:
-No strigi indexing (no attachments/non plaintext contents)
-No encryption handling (no indexing of encrypted content)

So this first version isn't feature complete, but should at least produce sane data.

Implementation of an accessible interface for MailWebView

Initial version of the interface, only has some functions of
the text interface

Created MessageViewerAccessibleWidgetFactory

REVIEW: 102410

implement mark all as read (for single feeds for now)

do not show a trayicon on startup if the user's preference is not to
have it

Add a checkbox to deselect the internal PSQL server.


FIXED-IN: 4.8.0

Get rid of the storePassword settings.

This simplifies things a lot, and fixes at least one bug.
Now password handling seems to work correctly in three cases:
- KWallet disabled in system settings
- KWallet enabled, but acccess denied
- KWallet enabled and access to it

The resource now tries to always load the password from wallet, which
gets rid of some corner-cases.

Get rid of the reset/layoutChanged storm

Advertisement:
Are you tired of loosing your selection in the message list when a
filter moves a mail from your inbox to its final destination? Are you
getting blind due to the favorite folders list flickering when a bulk of
emails are moved around? Are you seeing strange crashes and kmail
getting unresponsive when you move a big chunk of emails from one folder
to another? Then this patch is made for you!

Further explanations:
Don't use the itemMoved() signals in the ETM for items anymore. It is
the right solution long term, but short term it is not usable if all
the proxy models around the ETM don't support the feature properly.

Right now most (if not all) of our proxy models transform itemMoved()
signals into layoutChanged() signals, and for instance when that
reaches the message list it simply resets the whole message list model.

Such a reset then happen for each of the item moved, potentially
flooding the system with computations for threading, but also generating
flickers in the favorite folders list or loosing the selection for the
less dramatic cases.

Downside:
Because of the move speed up, you actually see the changes now, so the
mails progressively disappearing from the source collection to appear in
the destination collection. For this to disappear we need to compress
the changes in bulks, which is ongoing work. But at least now it's
faster and more usable.

EMFPLUS: A new infrastructure for parsing EMFPLUS records inside EMF files.

This is the first commit. It is still not connected to the rest of
the parser infrastructure.

Add the ability to use previews for frames with unsupported formats.

It will show the embedded objects in order of preference. If the
first object is not recognised, the new Unavail shape will be used,
which will then try the other options in turn, and for each attempt
to load a Picture shape, and upon success it will load the data and
disable the Picture shape's tools (to avoid confusion etc). Finally,
if no object is recognised, a question mark image is shown, to
indicate that the object is unknown.

Work done by Inge Wallin and Dan Leinir Turthra Jensen.

REVIEW:102512

feature: New menu to lock/unlock docks

Disable loading tutorial on startup

It's still available under the Help menu.

Add posibility to not duplicate markers

Use the addMarker to make sure we will not duplicate markers once duplication
checking is implemented.

Added GHNS for icc profiles
Category in the server is not yet created. Remaining things are done

starting videos from commandline works with KUniqueApplication

rework VPN status overlay icon

we need to trigger a repaint on activation/deactivation of a VPN connection,
otherwise, when using a wired connection, which does not receive
signalStrengthChanged signals and thus does not repaint every few seconds,
the VPN overlay will only be shown after a click on the NetworkManagement icon.

I guess this commit also improves performance, as we don't need to iterate
over all available VPN connections at every repaint, but only over the active
connections, and even that only when the tooltip is about to be shown.

This commit also makes an ancient hack, which originated due to not working
autorelaying of signals on our session bus service, a little less hacky

fixes a crash in plasma when kded is restarted

Add a new class to watch for the availability of krfb.

This is because the self contact capabilities do not advertise
the availability of krfb; they only advertise the availability of krdc.

Introduce Wicd dataengine.

Update the sidebar if a note is modified or created

Ideally, we should just use the ResourceWatcher instead of the
notewidget informing us of the changes, as some other application
could also be creating notes. But since nepomuklib doesn't
currently work with Nepomuk::Resource, we'll have to stick with
this method.

I'm using a QSortFilterProxyModel to keep the list of notes sorted.
It seems a bit of an overkil. Maybe I could use some kind of
priority queue?

Add a delete note button on the view

It appears with a timeout of 500 msecs. Actual note deletion still
hasn't been implemented.

First attempt at a sidebar

It's shit right now - Currently it's acting as the main widget, and
showing a list of all the possible widgets.

I've added a custom delegate but that looks horrible.

Create a NotesView

Now the sidebar is no longer a QListView, but it contains one.

This appraoch is better as I always wanted the side bar to have a lot
a of stuff like tag suggestions, and all.

Initial commit of untested NM API 0.9 patch for knetworkmanager
Base code and changes by Ilya Chernykh, NM 0.8 and below handled by Timothy Pearson
This does not affect NM 0.8 or below

Add an initial C++ port of the smartauthmon script to smartcardauth
This still relies heavily on system calls, but is already faster and has the potential to be more secure

kwin: Optimizing Blur Part III

This patch introduces some kind of damage propagation. In the old version of the blur effect
we had to repaint the whole window if sth behind the blurred area was damaged. The new texture
cache, which was introduced by the last patch, gives us the opportunity to only update parts of
blurred background texture. This means that the damaged area can only propagate with the speed
of the blurring radius per window layer.

REVIEW: 102665

Undo manager: Don't prepend() to a QVector which can contain many, many items.

Use a QList instead.
A QStack is a QVector, which has an O(n) prepend method.
Conceptually it is a stack, but not always used like one, so I
haven't renamed the typedefs, so the code looks a bit (more) confusing
now. Sorry about that.
I noticed that copying several 100k files slowed down to just a few
files a second and found that most of the time was spent in
QVector<KIO::BasicOperation>::prepend(), called from
KIO::CommandRecorder::slotCopyingDone().
Another problem is memory usage, which was a few hundred megabytes.
This patch does not fix memory usage because I have a lot of memory
and it's a difficult problem :)

Update the layout periodically when resizing the window

Up to now the layout has only been updated when the geometry has
not been changed for at least 300 ms. This has been changed so
that at least each 300 ms the layout is updated to give a better
visual feedback when resizing the window.

Improve schedule editor ui

Add possibility for more detailed control and info dependent on selected scheduler plugin.
Improve i18n tooltips and semantics.

Cache the size of the KoShapeGroup

This speeds up handling of groups with a lot of child shapes. This is especially noticable for
page changes in stage. Before the change it could take more then 10 seconds and it is not instantly.

REVIEW:102645

Improved performance of deleting items from data project

Previously each item was being deleted one by one. Now whole
ranges of items are removed at once (if possible i.e. the ranges are
continouos). This speeds up the process significantly.


FIXED-IN: 2.1.0

Improved performance of adding items to data project

Performance of adding items has been improved by refreshing model
after all items are added to model. Previously model (and thus GUI)
was being refreshed at each item added to it. For large item sets
(hundreds of files) that process have been taking a lot of time.

Hide password from debug output

Reviewed-by: David Edmundson
REVIEW: 102667

remove obsolete code from KRss. That is, most of it.

Single, plugin based akonadi-nepomukfeeder

This is a refactoring of the nepomukfeeder agents to a plugin based
architecture, so we can control the system load better in a single process.

With the new approach plugins can be written to index specific mimetypes.

Other changes include:
-No Strigi indexing code in the agent (this should go to the relevant plugin)
-No activity detection (the feeder shouldn't place any noticeable strain on the system anymore)
-Backport of a couple of changes which were applied to the kdepim copy
(error message, progress reporting, disable feeder config)
-The feeder will always only index what is in the local cache, and not download anything.

Set a program name to show a nicer name in the kwallet dialog, not just "Akonadi Resource".

Merge branch 'desktop-sharing'

Reviewed-by: David Edmundson

Cleanup NoteWidget + NoteEdit and disable automatic saving

The automatic saving when one starts typing is useful, as it
helps correctly set the nao:creator. But I can't seem to implement
it properly. So, for now it has been disabled.