Issue 174

Fix standard include dirs and macros under Windows
This patch enables Kdevelop to find the include directories and standard macros under Windows.

There is no "gcc -dM -E -" under Windows. So you need to do a hack to find the standard defined macros. Therefore a empty cpp file is compiled with a special commandline option "/Bx" to replace the compiler against a program which interpretes the output of the preprocessor.

For include directories Windows uses enviroment variables, so this patch processes them.

This patch has been written by Patrick Spenderin, thank you for that.

New: gluon_qtplayer crashes with not existing target

Sync config changes made while importing to the disk,
so that they're not lost if your computer doesn't work well.
(yep, can't be that kdevelop doesn't ;))

Replace the full word when completing paths.

Actually use right method name.

Sometimes I really miss static typing. This is the kind of issue that is
supposed to be caught by unit tests in dynamic languages, but the test
suite isn't close to that extensive yet.

This fixes bug 268072 (kdesrc-build without modules passed in fails).
There is a test case which simulates kdesrc-build without modules passed
in, but it doesn't simulate the entire sequence and so didn't catch this
error. My apologies.

Use proper help index naming as suggested by the following bug report.

Another round of strings/UI fixes to improve internal consistency as well as compliance with KDE HIG. I wanted to pull this through as long as we're
not translated. Changing strings becomes much more difficult after that.
If there's anything you don't like, just complain! If you notice something I have missed, either fix it yourself or inform me.

Fixing some bugs discovered when playing with a tail function implementation.

Legends and labels now properly handle curves with different units,
by putting them in the legend rather than in the axis label.

Apply patch by 'styx' to try to fix bug 258431. The patch only partially fixes the issue (see https://bugs.kde.org/show_bug.cgi?id=258431#c16 )

Fix NULL pointer crash in ScanParams::slotNewResolution()

Happens when looking up the current scan resolution if the
SANE_NAME_SCAN_RESOLUTION and SANE_NAME_SCAN_X_RESOLUTION option
names are not the same and the scanner only has the former. They were
equivalent in SANE 1.0.19, but in 1.0.20 they are different.

Fix off by one error on setting the search boundary limits
Fix researched by Christopher Reichert
Will backport to 4.6 if works fine in trunk for a while

Fix handling of unknown mimetypes.

If there are no associated apps, or the mimetype is wrong/unknown, then use
KRun to show an open-with dialog or determine the mimetype correctly.
Same logic as in KonqMainWindow.
My testcase was an attached MESSAGE/RFC822 (uppercase, so unknown) file in Jira.
(cherry picked from commit 528fb4202ff7d55d693b6f175322f2acf0d6e753)

Explicitly declare another variable

KCommentWidget: Don't interpret HTML-tags in comments

Thanks to Wolfgang Mader for the patch!

Add KAuth support to KWin process killer

See http://svn.reviewboard.kde.org/r/4460/
Thanks to Mikael Gerdin for the patch.


FIXED-IN: 4.7.0

Fix for connecting to FTP sites through HTTP proxy servers.


REVIEW:100814
FIXED-IN:4.6.2

Send the correct suspend type mode for suspendSession action.


FIXED-IN: 4.6.2

Fix stop() killing the list job even if another dirlister needs it.

Regression introduced by me in bef0bd3e3ff.
Symptom: "dolphin $HOME" showed up empty.

In the case of concurrent listings, I made the use of the cached items job
conditional (only created if there's anything to emit) so that we can join
the current listjob without killing it (updateDirectory) if it hasn't emitted
anything yet.
The unittest also uncovered inconsistencies in the emission of the cancelled
signal, now cacheditemsjob behaves like the listjob in this respect.

FIXED-IN: 4.6.2

(cherry picked from commit 51707e7154082b549216b8a8ecde73505302fadc)

fix & cleanup Workspace::setNumberOfDesktops()

encapsulate all relevant (vector resizing) code to
Workspace::setNumberOfDesktops and invoke it from
Workspace::loadDesktopSettigns, this will likely
fix

Unlink maximization state from actual geometry

Since clients with restricted geometry now cannot cover the entire screen by default it was necessary to unlink the state from the actual geometry to re-provide the "restore" feature.
The patch also extends the protection of the maximized state to unilateral maximizations (emacs issue)




review request https://git.reviewboard.kde.org/r/100606/

check that we have an item to query

Fix passing matrixes to the shaders

QMatrix4x4 accepts data in row-major order, but returns them in
column-major order, which is not documented and because of that
I expected them to be in row-major order.
This commit fixes it and rewrites the shaders to apply the matrix
multiplications in the right order.
REVIEW: 100759

validate deleted moving window in desktop grid effect

Remove reliance on a thread.

We don't need a thread when the event loop is present.
Listening for updates and persisting after job is done
is not the right thing for a KIOslave to do. Instead
this should be moved to some daemon.

Fix regression in charset parsing.

Implicit LWS is permitted between tokens in key/value pairs.

Add paths to qt plugins (from kstandarddirs) before qapp is created.

Necessary for apps started through d-bus activation, so that the kde widget
style is loaded. The QCoreApp::instance test was "so that qt.conf is read" (db852684)
but the whole idea of 9d7eb8f7 is that this doesn't rely on the qt config files anymore.

FIXED-IN: 4.6.2

(cherry picked from commit 979a19afe93d1e4a5be684c1139b6ab242e6b9b6)

Usability fixes

Make the edit area disabled until the user selects he wants custom
positions.
Its a source of confusion if the user can edit the positions but doesn't
see anything happen in his decorations upon apply. This gives immediate
feedback that the checkbox should be enabled.

Conflicts:

kwin/kcmkwin/kwindecoration/buttonsconfigdialog.cpp

Fix some search issues

- Always merge the query from the "Find:"-field with the current facet-query. This assures that the facet-query never gets overwritten if the user changes the "Find:"-field.
- Always trigger a searching if the text of the "Find:"-field has been cleared.

Still a lot of usability issues are left, but this is at least a first step...

Fix GL error when do texture from pixmap

This problem happens on Freescale imx51 ARM platform

Use the correct ContentMimeTypes when creating a collection.

When displaying recurring to-dos, properly handle daylight savings shifts.

When dissociating a recurrence, which is an "add" and a "change", only add if the change was successful.

Fixes duplicate incidence problem in issue4096, msg27582.
MERGE: trunk

First ask for KWallet, and then show up the splash screen ;)
Also, Splash screens are closed by clicking on them!

Don't draw recurring to-dos at the wrong day.

To-dos due at 00h00 should be drawn at the previous day and ending at 23h59.
For recurring to-dos we don't have this implemented so i'm disabling this
s/QTime(0,0)/QTime(23,59) code. Otherwise we it woul be drawn at 23h50 of the
current date instead of the previous.

Fixing errors in google docs

Bug fixes in the saving of changes in ODF 1.2 format.

fix unclipping shapes

correctly determine the transformation needed to transform the clip path
shapes from their original position at the time the clip path was created
to the current positon of the clip path. both transformation changes of
the clipped shape as well as any resizing applied to the clipped shape
have to be used to create the resulting transformation.

DOC: Improved processing of annotations (comments).

* Wrong CPs into the Comment Document were parsed.
* Wrong offset to the Comment Document was used.
* Process comment reference characters only in the Main Document

save angle and scale in predefined brushes

: Export CSV not unique identifier

RCPS scheduling

Mainly improve handling of time constraints.
Fix incorrect resource allocation calculation.
Handle FixedInterval tasks correctly.

librcps:
* Change fitness from a simple int to struct with group and weight
to facilitate optimizing on different criteria, like
time constraints, project target time, duration, cost, etc.
* Add fitness callbacks to handle the new fitness calculation possibilities.

Disable View management (close-tab and new-tab) when using PulseAudio
Fix "Can't add new tab when I remove all tabs"

Emergency fix for Pulse support - CaptureCategory issues

Ask delete confirmation for all contents of folder.

Apply base download dir setting.

Use persistentindexes for deleteAction.

While deleting groups the indexes got out of date. QPersistentModelIndex prevents that.

Fix DVD times being really really wrong. Turns out that rsndvdbin doesn't send out duration messages when the duration changes.

Don't overwrite playlist tracks.

When a playlist was not lazy loaded yet it would be possible to overwrite the tracklist.

Be more lenient towards torrents with wrong announce-list field

This commit fix bug 266837 about the (in)ability to close rekonq,
but IT CHANGES Application::mainWindow() behavior to an unsafe mode.
It is now basically allowed to return a null pointer, leading to other possible crashes.

Some days of testing since now, for this.

Close the channel when the tab is closed

Sanitize the Networking, Wireless and Wwan checkbox handling
according to Dan Williams comments in

Fix crash on "s" type :D
This crash was reported with this bt (reduced here):

Hence, the fix. Just checking if the QBitArray is empty.

Fail gracefully if required QCA features are not available at runtime.

Konversation would crash during Diffie-Hellman key exchange or Blow-
fish decryption/encryption if QCA was unable to provide these features;
usually because the qca-ossl provider plugin isn't installed. Now it
will show helpful error messages instead.

Thanks to Johannes Huber for the report.

Checking for the empty string causing Rohan & Pano's bug...
Next test



By the way, are you sure this is yet happening? I can no more reproduce it since the first fix...

Return false for null mimetypes

Replace cmd script with a new version written in Python.

- Fixes a bunch of bugs in the old Perl version, such as claiming
a command does not exist when it returned no output or only empty
lines (now an appropriate info message gets shown to the user in
that case), as well as generating faulty D-Bus calls when trying
to forward empty output lines to Konversation (output lines con-
taining only whitespace are no longer forwarded).
- stderr output is now shown to the user (not sent to the server,
of course) instead of being silently swallowed.
- Trailing whitespace is now stripped from output lines.

Another one in the "fixing the dangerous commit" series.
This one is for crashes on restart after a bug...

This is a first commit fixing the "dangerous" changes introduced to fix
bug 266837.
We basically have now a mainWindow() slot widely used in rekonq, that can return
a null pointer.

Checking the code again to ensure this is the unique bug caused by the change...

Removed ::roundMask method due to removed GdkPixmap.

TODO: reimplement with gtk+3.0

Fix bug #268239

Fix misplaced space character upon inserting a description list.

Patch by Mark Morschhäuser.

Fix bug #268236

Fix button states on the device selection dialog

When moving a file from/to a temporary location, also update the file access/modification times.

When PHP moves a file across filesystem boundaries, it does not update the access/modification times.
Thus do it manually so that this information is not lost.

Ask user whether to clean the repository after multiple crashes.

REVIEW: 100822.

Guessing of types for attribute access autocompletion works!

Pretty easy and straightforward, the current line which should be
autocompleted is cut so it doesn't have syntax errors, then thrown into
the parser, and the expression visitor can then guess the type. Almost no
extra code necessary so far (just restructuring).

Guessing types for autocompletion works for member functions now, too

... for example something like foo.bar().?? could theoretically be
autocompleted now, given we know the return type of the bar() method
(which we almost never do currently).

QmlDesigner.qmlPuppet: allows specifying alternative to qmlpuppet

This allows the user to specify another qmlpuppet then the one
delivered with QtCreator. For example a pupper build with mingw.

The enviroment variable QTCREATOR_QMLPUPPET_PATH is used.

Reviewed-by: Kai Koehne

FEATURE:
Add a data source config dialog option to the change file dialog, to allow configuration options to
be set for the new data file.

Use QToolTip for showing some info for object under cursor.
Much simpler than previous version also prettier and more informative

MapWizard: have metadata page rather than legend page appear after map source page

pros:
* map source is visualized in metadata page -> better feedback
* legend data is optional, and therefore not as critical as metadata

Implement support for overlays to operate on multiple images (= on current selection).
Provided as an opt-in solution for each overlay, with as much code shared as possible.

Implement operation on all selected images for the rating on rotation overlay.
On the way, clean up code duplication mostly in the rotate overlays.

Question: Now, one click can be very powerful and edit a large number of images.
Should we add a confirmation KMessageBox if the number of edited images exceeds
a constant (n = 5, 10?)?

Apply local patch which creates an 'Configure Konsole' menu and dialog.

Some profile options don't actually belong to profiles. They should be
in the main Konsole settings. This is the first step is trying to
seperate the options. I only did 1 option for this patch.
At some time it may be possible to put the profile dialog inside this
main settings dialog.

mobile network dataengine:
* update the source when a property changes.
* allow to specify a different modem.

Add support for offsets to the temperature applet.

Useful when your temperature sensors are off by some degrees and
you want to calibrate your readings.

Added a new data engine that can be used to get information about the current phone profile (ex: silent, general, etc...).

Add support for updated_time for events.

Eventually this can be used to only update events that really changed,
instead of unconditionally updating all events every sync.

Show Nepomuk specific metadata

Adjust the protocol of KFileMetaDataReader to stream Nepomuk::Variant values.

Added 2 new mobile phone applets:
* Mobile Network Operator: displays mobile network operator.
* Mobile Network Status: displays signal type and strength.

Show participants to events.

We now show a list of participants to event including status (accepted,
declined, not sure or not responded). Email is set tolid
since we do not know it.

Replies are also added

MessageID, In-Reply-To and References are set correctly so KMail does
threading!

Add option to use KCal instead of KCalCore, to work with KDEPIM 4.4.x

The option is enabled by default.

Messages show up in akonadiconsole now

No replies yet and a lot of things have to be extended but the general concept
is working!

TODO:
* from who is the message
* to whom is the message
* date (how to get the date of the first message??)
* reference (so we get threading)
* and probably much more!

Implement reverse geo coding.

That is, retrieving the address for a selected map position. It doesn't
do anything with the data yet, though.

It uses a class that's exported by Marble, but the header is not installed,
so we need our own copy for now.

First attempt on Marble integration into the contact editor.

It's a quick hack but already way better than the previous tiny world
image.

Show participants to events.

We now show a list of participants to event including status (accepted,
declined, not sure or not responded). Email is set tolid
since we do not know it.

List of message threads is retrieved

The list of message threads is retrieved which will allow for easy checking if
a thread needs updating.

Forms: re-add foreground and background color properties for widgets

Undo/redo works too.

Implemented a new bounded tile data pooler

(!) Important notice: it would be more safe to perform 'make clean'
in krita/image/ directory if you decide to switch to this branch

Well, this commit implements a new memory-bounded pooler thread. The
configuration of the pool is done using a special option
"memoryPoolLimitPercent". It sets the percent of the memory, that you
would like to spend on the tiles pool. Please notice, that this value
is included in hard/soft limits as well. That means Krita will never
take more than Hard Limit memory on tiles and clones together.

This implementation performs not full power yet. I haven't done any
optimizations, so the highest efficiency it can reach now is 50% of
tiles pre-clone hits. I could get this numbers on gradient tool and on
very short brush strokes. It is quite curious, but the longer stroke
you do, the less efficient the pooler is. I hope i'll get it fixed
soon.

The applet for banks is now functional

Fix detection of mounts of NFS & SMB Shares

NFS & SMB Mounts do not emit an accessibilityChanged signal to
Solid::DeviceNotifier. This commit implement a timer to monitor
mounts. But only works with KDE 4.5 because of use of KMountPoint.

Improve AvCapture so both devices can be set at the same time

Shoot the MainMenu && add SubtitleMenu

MainMenu, being ambiguous, is now dead and RootMenu takes its place as
supreme item. Apparently Root and Title are sometimes a bit messy (e.g.
on my Interstella DVD root is grossly depending on the backened, and
they all get it wrong...).

Also add SubtitleMenua as those are sometimes not part of the AudioMenu
(also reflect this in the documentation).
AngleMenu is now last as no one supports that junk anyway.

Add dvd menu navigation interface

Respect the users Download path set in KDE System Settings

Implement support for channel specific highlights

Support to read/save the plasmoid's state (Networking, Wireless, Wwan).
I had to use the same QTimer::singleShot hack I use with Wwan with the
other two to make sure NetworkManager will respect the plasmoid's
checkboxs status.

By the way, the Wireless checkbox is not working properly because of
a missing wirelesHardwareEnabledChanged emit in Solid's NM backend.
I already commited the fix to KDE/4.6 branch:
https://projects.kde.org/projects/kde/kdebase/kde-workspace/repository/revisions/2ec6533494a23164ee9c2de4ab65d5fa372aff5c

Add search modes for the Search Panel

The search panel must get a hint whether clicking on the facets should result in searching everywhere or from the current folder. It is not sufficient to check the search-settings of the "Find:"-box, as when the "Find:"-box is invisible there is no hint for the user what kind of searching is done and the setting must be ignored.

implemented window background for mainwindow, toolbars, and menubars.

33 to go

12 to go

Create method to load just note metadata details without parsing the whole note.

Work as a starting point to convert system tray to a plasmoid.

Sync default keyboard shortcuts with Konsole.

initial port of the remember the milk plasmoid to pure qml+javascript. Currently it is just able to show TODO-items in a list and show different lists in a tabbar. Editing the tasks is not possible yet. Also the authentication isn't saved to the config-file and thus needs to be redone on each restart.

Add a directory selection button (and the corresponding dialog) which allows it to
select the location of the TeX documentation in a more comfortable way.

Patch by Felix Mauch

Made the QueryModel async

Make everything asynchronous!

feature: New button in "operation" page to freeze/unfreeze all fields with text

Use KSpeech to speak computer's moves.

Keep better track of dragged and selected pieces.

Use libwww-perl instead of Net::HTTP/Net::FTP

Sorry about changing dependencies, but LWP seems to be the standard Perl
solution for downloading files, so it's likely to be more prevalent
anyways. Neither group of dependencies appears to be part of core Perl
so I'm having to pick either way.

Picking this way allows for a large reduction in code usage while still
properly handling redirects.

Significantly improve automatic legend names.
I think I now do a reasonable job with all cases, except the case where
units are not consistent between all relations in a plot.

Please look for troubles.

Somehow the KDebug include is not working correctly, we get a lot of warnings
during compile. Using "kdebug.h" works, the AREA_CODE for kDebug can be read
and no compile warnings are visible.

Greatly improve the performance for the metadataselector class. By adding all
treeitems one-by-one the treewidget seems to trigger a sorting operation.
Adding many items like in the "Makernotes" tab resolves in a poor performance,
on my computer it takes nearly 10 seconds to add all the items, on a friends
computer it's even worse.

Adding the toplevel items parentless to a QList helps to improve the
construction of this tab.
It only takes up to one second now.

Details view optimization

Constructing a KColorScheme object is very expensive because of a number
of tint computations. When scrolling a big list more than 30 % of the
time was spent here. Instead, we can precompute and store the inactive
text color. (see https://git.reviewboard.kde.org/r/100826/)

Thanks to Samuel Rødal for the patch!

Set the modified and created dates of events.

This is an attempt to keep the payload of event items always the same,
after every sync.
However, KCal seems to set DTSTAMP to the current time on each sync,
so the payload actually differs. Apparently nothing can be done about
this, short of changing KCal code.

Prevent unnecessary reloading of KDirLister on startup

When Dolphin is started with a directory as argument unnecessary reload operations of KDirListers are done. The patch improves this by just letting the DolphinView internally do a loading as soon as it gets visible (and without reloading). Also in case if only one directory is passed as argument the current tab gets reused instead of creating new tabs and removing the current tab.

Thanks a lot to David Faure for pointing out this inefficient handling.

Run photoJobs in parallel

By running all the photoJobs in parallel the friend syncing gains a huge
speedup! There is some serialisation nessessarry but the main advantage is that
all the pictures are fetched in parrallel.

make compliant with XDG standard

Instead of synchronizing the whole account tree on each data changed take advantage of the new, more granular, signals and perform the appropriate modifications in the model's data after it has been loaded once.
Also create a separate instance for the institutions model. This change has the following rationale. Mixing the two hierarchies in the same model (by account type and institution) would make it difficult to implement changes since depending on the changes it would imply structural modifications in only one instance of the same account. So for example changing the institution means a structural change in the institution hierarchy but no change in the accounts by type hierarchy. Separating the two hierarchies means that one account can be found only once in the model and it also simplifies the implementation of the modifications slots.

REVIEW: 6589

Significant improvements to state handling. Needs tons more testing, especially wrt network streams

Cleanup layout syncing and apply HIG layout for all options

improve the names of the vcs icons

adopt a format like "vcs-<action>-<vcstype>-<vendor>",
so there is no clash with other potential icons, and at the same time it could open the door to possible generic (non-vendor) icon names

also, move them to the hicolor theme, so they will show up also when using an icon theme different than oxygen (but keeping the possibility to be overridden by the current theme)

Start making the Maemo support more generic.

This includes:
- decoupling deploy configurations from targets (Reviewed-by: dt)
- adding a "Generic Linux" device type
- splitting up the Maemo deployment step into small pieces that
can be combined in different ways (and much more easily maintained)
- adding a new version handler for pro.user files
(Reviewed-by: Tobias Hunger)

Also:
- Add and use an SSH manager class for easier connection sharing.
- Make the SSH connection parameters a fixed attribute of the connection.

JAVA implementation for QT Image Capture component.

Added a template Ontology

Many Nepomuk apps need to write their own ontologies or extend existing ontologies. This is a template ontology to get one started.

Revise how test suite is run.

Fixing the test suite for my recent refactorings ended up being quite
the trip through the looking glass.

I ended up making the run-test phase not enabled by default due to
always filtering it out in the test suite. So naturally I just add it
back in if we're using the run-tests option. But naturally *that* change
warrants test suite coverage, and testing it out gave me error about
finding main::list_has, since list_has is really test::list_has when the
test suite is being run.

Using Perl's caller function allows me to reset the package to main at
the start of kdesrc-build and still detect when it's being activated as
part of the test suite, but then the functions in the test suite don't
work until I switch *its* package to match.

It all does seem to work now though.

Start on revamp of kdesrc-build code strcture.

So what I've done so far is to introduce a "BuildContext" class, since
it doesn't make sense for a simple Module class to also track the global
module list. This also makes the phase handling (also split out to a
different list to avoid code duplication) a little more normal.

Now the only global state in Module is related to the source of where
modules to be built was established (config or cmdline), which probably
needs to be moved as well.

In the long term much more will be modularized (even though I'll
probably continue to leave it all within one file).

Make sound output the default instead of voice navigation - so far we don't have a default voice to ship.

Porting the account applet to plasma, using dataengine. Still need some work

Move text layout out from kotext and textshape into a new library called
textlayout.
The actual layout is disabled, and Words is also disabled, but all
applications should compile, and it should be possible to load and edit text.
Though naturally without layout you will not be able to see anything yet.

Seems to be successfully compiling loads of stuff now...
Add XPLATFORM_ANDROID to configure, used it for OpenSSL compile.test.
Add ANDROID_GLES2_LFLAGS (-Wl,–allow-shlib-undefined) and pass it through to (all!) OpenGLES2 compile-tests.
Accept -Wl* as input to compile.test and pass it on in LFLAGS.
Back to android-9, of course, 10 doesn't exist from an NDK perspective and 11 isn't out yet.