|Commits||1572 by 145 developers|
|Bugs Opened||775 in the last 7 days|
|Bugs Closed||680 in the last 7 days|
|Aaron J. Seigo||
|André Marcelo Alvarenga||
Internationalization (i18n) Status
|Aaron J. Seigo||
|kde martin-graesslin com||
|Hugo Pereira Da Costa||
There are 195 selections this week
only take encountered declarations into account when checking for redeclarations
Fix bugs and modify behaviour of filter action
* delete the QSortFilterProxyModel when we remove an output
* when we create a new output -> crash
Now when we switch between outputs, the filter is change and
the output updated in accordingly.
Search pattern list is no more deleted; added icon to search button in grepdialog
Fixed context ranges
As the python parser does not provide "end" information, we need to go up the parent chain when encountering, for example, a statement in a function definition, and adjust the ending positions of all parent nodes (especially the function definition).
Otherwise, the context cannot be created correctly. This has been a problem before, but reappeared when we switched to the new python parser.
It should be gone now, anyway, together with the ugly "body.last()", "body.first()" hacks (the FunctionDefinitionAst node now has proper start and end attribtutes set).
-Fix up manually supressed axis labels. Conceptually, manually supressed axis labels do not
work well with shared axis boxes, so supressed axis label settings are ignored while
in a shared axis box.
-Allow data dialogs to not have 'edit multiple' in the UI, and then remove it from the
vector dialog. The 'edit multiple' capabilities of the vector dialog are achievable in
the change file dialog and the change sample range dialog, so perhaps edit multiple is
never needed for the vector dialog.
Work around gpsmm parsing gpsd output (C locale) with current locale.
Allow using absolute source directories in texture layers in .dgml files.
Extend .dgml by a tileSize tag (width and height attribute), usable inside textureLayer tags. Such .dgml files are backwards compatible.
Handle 'base tile missing' case more gracefully and return an empty image instead of triggering an assertion.
Move the default tile size (675) to global.h and replace hardcoded default tile sizes with it.
Work around gpsmm parsing gpsd output (C locale) with current locale.
Backport of commit 1217540.
Delay by 500ms before jumping after a search text is input
Accept slight pixel movements for detecting mouse clicks
Should help with bug 263314
Synchronize selection between the icon view and the thumbnail bar.
This allows multiple selection in the thumbnail bar with a meaning: all actions
are carried out as if selected in the normal icon view.
Drop using the imageActivated signal of thumbbar (with the same as clicked() anyway),
simply use the current image.
This also fixes the problem that the arrow keys in the thumbbar would move
the selection, but not update the shown preview (may be a bug report somewhere)
- fix AlbumTreeView's expandMatches so that a DirectMatch does not prevent
recursion, in case there are additional ChildMatches.
- clear search bar when setting the assigned tags filter
(the assigned tags filter was already cleared when changing the search
bar, now it is symmetric)
Check available memory before attempting large allocations.
Missing commit part
Do not access zero pointer.
Cropping pictures that use a small palette (e.g. 256 entries) works now.
QImage::copy(QRect) also preserves the EXIF-Tag.
Ensure that the current index remains visible when the view's size changes.
Especially useful when changing the dock position of the thumbbar.
It can be annoying if a sidebar is shown/hidden and the current index was not
previously visible (scrolling without selecting). needs some polish for this case.
Show error message if imlementation is not supported.
Implement a fallback mechanism for OpenCV detection: If FIND_PACKAGE(OpenCV) can not find OpenCV, try to find it again using another version of FindOpenCV.cmake which is shipped with libkface and can still find OpenCV on some systems.
Tested under debian sid:
Normal installation of OpenCV via apt-get: OpenCV is found at the first try.
Remove /usr/share/opencv/OpenCVConfig.cmake: First try fails, but the FindOpenCV.cmake shipped with libkface still finds OpenCV at the second try.
If no OpenCV is installed at all, CMake stops after the second attempt, as it should be.
Do not restart playing if currently playing the same ImageInfo
No need to emit curvesChanged signal from mouseReleased.
It's sufficient to release it from mousePress and mouseMove.
Use update() instead of repaint().
Only use left button click for rating widget
Do not use time_t when calculating average date, that works only on a limited scale.
Use QDate's Julian Day to get a much wider range.
Fix check: Do not refuse to rename if relative path is identical, but albumRootId changed.
Remove stale album from destination location, not source location.
Fix crash with invalid node ids, cleanup
Don't cache files that are bigger than the maximum cache size.
NOTE: this fix only works for cases where the content size is known and
as such further improvements are needed in the future. Until then this
should fix all the problems reported in BR 244215.
Fix ksmserver crash in ~FadeEffect() on logout
Use _mm_free() for image->data because it was allocated with _mm_alloc()
in the constructor. Otherwise this might lead to a crash in XDestroyImage().
The crash is 100% reproducible on kfreebsd-amd64.
Patch by Modestas Vainius
Correct Enlarge Font shortcut such that it actually works.
Handle when tabs are closed via the close tab button on tab bar.
- make sure to not realloc ScanDirVector to avoid a dangling pointer in ScanItem which can lead to a crash
- added some const + QLatin1Char
Disregard all errors received on valid redirects.
Fixes BR# 263817
Improve debug message when receiving errors.
Forgot to emit changed(false), which caused saving to happen, but notification to the user to fail.
combo for elements within the webpage
This also adds
- a bit of sugar on top of KGraphicsWebSlice APIs,
- refresh should really be a signal,
- by making the frame accessible, we can actually find out what to slice,
- also fix some docs.
- add a reload button to show the newly selected webpage, in order to select the frame without reloading the complete dialogue
Stop using the active modus.
Disable tab close methods until the issue of menu being trashed can be resolved.
single click opens the config
activated() is checked against KGlobalSettings::singleClick(), clicked() isn't.
do not sync items after reacting to external changes
This hook prevents cyclic updating, which happened when Lion Mail was
displaying an email which got deleted from KMail2.
We simply mark the deleted item in that case, and it'll just go away,
without being able to be "undeleted".
avoid (non-critical) warnings
Rounds value instead of truncating it. This fix a problem with increasing brightness key not working sometimes.
abstractItem can be null
bounds check pos
set status when menu hides
Disable the Special Dates and Planner plugins by default.
A new option is introduced for kdepim 4.4.10: KONTACT_ENABLE_MIXEDMODE_SUMMARY_PLUGINS (default: disabled)
This will prevent Kontact from starting twice when restoring the KDE session.
(and probably a few other Kontact bugs)
DO NOT BACKPORT. AFAIK, kontact from 4.6 is not affected.
Hide 'mailto' icon in incidence viewer in MOBILE_UI mode
Fix crash when changing to monthview.
Sometimes drawBackground is called before reloadIncidences(). This used to be bespin especific
but now i reproduced it with oxygen.
Fix massive memory use when opening the bulk action screen,
For some weird reason, QML eats a huge amount of memory when the item model
is set right in the BulkActionComponent file, but not when set in onLoaded
after loading has been completed. No the extra memory use is barely
Fix deletion of invitation mail after calendar action
Fix deletion of invitation mail after it has been added
to the calendar in kmail-mobile.
Fix calendar display after maintaince mode
Clear the bulk action quick search filter when leaving the
maintaince mode, so that all events are shown in the calendar
and not only the filtered ones.
Fix drag and drop on category
If you drop a category on categoryRoot, this create an empty category
and drop your category into. this code fix the problem and the category
are drop at the good place.
Fixed the mouse-click handling in message viewer
The coordinate mapping code in the MessageViewItem::simulateMouseClick() reimplementation was horribly wrong.
Now we use the correct implementation from the base class.
Thanks to Volker and Andy for brain storming :)
Switch off scrollbars in MOBILE_UI mode
Fixed canvas rotation and improved zooming.
The canvas now won't shift away anymore while rotating and the focus of the
zoom now is in the center of the canvas widget instead of the top-left corner.
However there are still a few things to fix.
Now there are some scrolling problems regarding the zooming and the
"Fit Page" and "Fit Page Width" zoom options don't work properly anymore.
Fix handling of top and bottom margins for Word95 documents.
The dyaTop and dyaBottom types are YAS-datatypes which are defined as signed 16 bit integers.
A negative value means minimal top-value and a positive value means exact top-value.
Since we cannot handle that in ODF (and the minimal-part was removed in >=Word97 anyways), we just use the absolute value.
Commit the porting of the DOCX table logic to the usage of the newer classes.
Finally our tables are much more accurate.
Change the way we read the current table style, get the style name rather than the style out of the available styles.
This way we can much better be sure we are doing the right thing depending on the context.
1) patches to csvprocessing.cpp referenced in http://svn.reviewboard.kde.org/r/6394/.
2) patch to csvprocessing.cpp to remove spurious chars. in msg.
3) patches to redefinedlg.cpp to avoid display of un-necessary and confusing dialog appearing when importing reinvestdividend items, which involve no checking account.
fixes to connection point loading and saving
do not convert connection point position to shape coordinates when align is given
properly save connection point position when align is set
Rewritten to output ODF rather than old-kword format.
This fixes the following issues:
- runs of text are no longer turned into paragraphs
- text formatting is actually applied now (bold, italic, font size etc.)
- empty paragraphs are now preserved
restrict conection point positions to bounding box of shape
this seems to finally fix bug 251529
Compile and link f-office again.
I had to merge koabstraction and f-office together cause it seems the design of koabstraction is borg'd in a way that doesn't allow an other fix beside rewriting to use proper oop or merging them both together again.
Sorry, but better something that compiles and doesn't have the perfect design then something that has the perfect design but doesn't compile.
Please fix it yourself if you have a better idea, thanks :)
Connection points are no longer referenced by index but by their id.
This was required as ODF specifies 4 standard connection points with
reserved ids from 0-3 and custom connection points with ids > 3 .
When loading connection points they may have ids that do not relate
to their index, so things would go wrong when trying to create connections
saved to files.
A rather basic support for loading connection points was added along
with some supporting functions to check if a connection point exists or
to get a specific connection point position.
PPT: Include the information from the TextRulerAtom while processing the text content.
* Fixes the horizontal positioning of lists and paragraphs, TextRulerAtom was required.
* Fixes the problem with incorrect level information in list-level-style-* and
nesting of <text-list> elements.
Forward port r1217597.
Also supress screen power management when watching a film.
Please note that this currently does not seem to work on 4.6 as powerdevil does not seem to actually trigger the appropriate action.
Prevent crash on exit when using Phonon-VLC.
Thanks to Harald Sitter (apachelogger) for solving this riddle.
VLC requires that XInitThreads() is called, so we are doing that now, but only on X11 systems (if my CMake foo is correct). Please test.
Remove phantom tracks
More non-collection album names and special handling for classical album artist
If a cached suggestion is older than one week, it is downloaded again, allowing cache
to be updated.
Reviewed and merged by me :)
Move data file jobs are now done one after the other
prevent annoying popups when typing 'about:' addresses
This evening I started working again on this bug.
I did all the tests against XSS and it seems ok.
Hope it's true.
Update view tab when switching to it, fixes data scan widget showing in the wrong place
libkyahoo: Apply some fixes related to sending and receiving files.
From the ReviewBoard request:
a) protocol (as reading from various places) suggests that first a
HEAD and then GET should be done. Current code was asking them in
b) fixed the way sending the header was done.
a) changed to an async method of doing the send
b) send buffer size is dynamic to try to send as much as possible
c) ... various - made it work.
Maybe fixes also #194833 although I'm not very sure what is that one
Original patch by Cristi Posoiu <cristi.posoiu AT gmail>, thanks!
Cannot backport due to the string additions.
Use libjasper instead of relying on the external jasper program.
From the original ReviewBoard description:
I saw that for yahoo - webcam images are converted by running an
external program. Personally I am not ok with this since running an
external program every 0.X seconds is a bit expensive. Also, not to
mention using disk files (at least pushing directly into the program
and reading its output would have been better).
Also - I saw that inviteWebcam action was checking for presence of
jasper program - but requestWebcam should also have done that. Not
to mention that if a yahoo contact just invites you to see his cam,
you might get a feedback of missing program after accepting to
invitation. Not necessarily a big deal, though.
Original patch by Cristi Posoiu <cristi.posoiu AT gmail>, thanks!
This commit is a forward-port of r1217895 to the 4.6 branch.
Fixed rendering of GtkComboBox
- removed all modifications of CellRenderers layout.
- set xthickness and ythickness of GtkComboBox's Frame to 4 (as opposed to 0)
- manually enlarge the GdkWindow of the GtkComboBox's button, to cover the dead area created by the Frame's margin.
- moved rendering of antialiased corners from Oxygen::ShadowCache to Oxygen::Client.
- simplified caching of ShadowCache accordingly. Notably, it should make it easier to pass the shadows to kwin.
- removed unnecessary QPainter::save()/restore() calls
- set painter AntiAliasing render hint earlier in Client::paintEvent
- modified drawing of drop target rect accordingly.
- oh. And fixed krazy warnings.
removed all black-listed applications.
New ARGB handling should now be safe for everyone.
Keep the list in the meanwhile in case the assertion above is not true.
(re)-Fixed panel applet background rendering
- renamed gtk_widget_is_panel_applet into gtk_widget_is_applet.
- simplified method to check both widget and parent at once
- moved (broken) call from Oxygen::Style::renderWindowBackground to Oxygen::StyleWrapper, where needed.
Revert "only connect to worker signals when we've requested it" changes. It's too difficult to determine when to connect, causing a ton of regressions.
This will just have to be addressed in QApt2, when the time is right for that. In the meantime, this issue is just a bit annoying.
Update the board after a move.
This way, removed pieces show up after undoing a capture move.
find more hints to solve more games
move whole lines too
allow pointless moving of kings if redeals are ahead, grandf is a strange game
freecell was never a game where empty piles allowed only kings.
I wonder how many games still could be solved
klondike 3 is not as easy with automoves as klondike 1
gypsy: don't ignore valid moves, but give them low prio
Added search history support
This support is still WIP !
New icons for forward one screen, back one screen, pause, read to end and count from end.
I think the latter two are muuuuch more intuitive like this, I hope everybody will agree.
SVG sources included, based on dialog-information.svg from oxygen, but largely simplified because there were aliasing problems around the edges at such small sizes when using all the gradients and shadows of the original icons.
There is more work left on icons:
- cleanup old icons which we no longer use
- make a couple new ones, right now export vectors to disk, export and log entry all have the same...
More new icons and some slight improvements to the menus:
- save vectors to disk + icon
- log entry
- export as image
SVG sources included, coming from oxygen except the vectors.
Files used: text-x-log (mimetypes), drive-harddisk (devices/small), resource-calendar-insert (actions), image-x-generic (mimetypes/small).
Oxygen icons are just great!
Delete folders, rename them, create new ones, delete bookmarks and edit bookmarks (including moving to other folders).
Also adds support for importing .kml files.
The only missing thing -- which I plan to make a junior job -- is drag & drop support.
(Internal move to change the order of folders and bookmarks inside folders. Maybe also D&D between views to support dragging bookmarks quickly to other folders.)
Squashed patch of a local git branch:
- Support importing bookmarks from .kml files. When conflicts arise (imported bookmark already exists), the user is asked whether to skip the new or replace the existing bookmark.
- NewFolderInfoDialog is a misleading name. Rename it NewBookmarkFolderDialog.
- Rename NameDialog to NewBookmarkFolderDialog, the former is too generic. Fix broken license header in NewBookmarkFolderDialog.h. Make it possible to use the dialog without changing existing bookmarks directly. Use a ButtonBox in the .ui to get a consistent interface.
- New bookmark folders can be created within the bookmark manager.
- Deletion of individual folders and bookmarks. Use the first folder as default selection.
- Use the existing dialogs to edit bookmark folders and bookmarks in the bookmark manager.
- Rename BookmarkInfoDialog to EditBookmarkDialog which describes its purpose better.
Position Marker trail by Daniel Marth
Adds a trail to the position marker.
Add KMemoryInfo to access information about system memory.
This class was initially written by Pino for inclusion to kdelibs,
but rejected for various reasons.
The API of this version is changed pretty much to my taste, the backend is replaced with code copied from libstatgrab, which also supports various Unixes in addition to Window and Linux.
Note: only tested on Linux so far.
Pretty obviously, this is stuff for kdelibs, but as we saw, it did not find its way there.
Simplified and improved the ad blocking code
move all launcher config reading/saving into GroupManager
GroupManager can now have a config group provided by a subclass;
a siple set/get method pair wasn't enough as Applet::config() can
change without notice (principle: Applets should never hang on to
the results of config(), but call it, use it, then discard the value
upon returning from the method it is used in)
* this is all in one neat, centralized place allowing launcher config
to be used amongst multiple implementations
* less code duplication just to satisfy config
* the launcher[Added|Removed] signals are now gone, which gets rid of the
somewhat ugly implementation detail that there were multiple ways to
find out when a launcher was added, some more direct than others
also, adds a configChanged signal which allows launchers to be saved
immediately to the config, safe(r) against interuptions / crashes
- Added a new class StopInfo to store data for a single stop suggestion, instead of separate lists of stop names, ids and weights.
- Added new TimetableInformation enumerables StopCity and StopCountryCode. To be used with stop suggestions.
- Vehicle types can now also be set in scripts using the integer value of the enumerable. There are now also more strings to be set as vehicle type such as "regional" for regional trains.
- Renamed VehicleType enumerables for trains, eg. from "TrainInterCityExpress" to "HighSpeedTrain".
- Accessors can now tell the data engine to first download a session key or stop ID before requesting departures/stops. The session key / stop ID is then used for the request.
- Added a TimetableInformation enumerable "IsNightLine", to indicate departures with night lines. But this information isn't yet used.
- The default vehicle type setting is now also used for scripts.
- all: Added changelogs and short author to the XMLs.
- at_oebb: Wrote a script for at_oebb and updated URLs and parsing. It can now also parse route information, journey news, platform and delays.
- at_oebb: Better stop suggestion source, faster by requiring less data to be transfered, added stop weights.
- ch_sbb: Updated to work for journeys again. The problem was changed HTML code from the SBB service provider.
- ch_sbb: Added new functions. Parsing of routes of journeys with departure/arrival times, types of the used vehicles, line string and platforms). Parse capacity information and add it to the journey news.
- ch_sbb: Fixed to show all departures, not only every second.. This was a one character fix ;)
- cz_idnes, sk_atlas: Wrote scripts for cz_idnes and sk_atlas, replacing the regular expressions in the XMLs. They can now also get stop suggestions by using an ASMX web service (with POST requests), which needs a session key. The session key is read from the service provider home page.
- cz_idnes: Fixed, didn't show new departures after some time.
- de_db: Date values are now correct for de_db, also if departures for a date other than today are requested.
- de_db: The de_db accessor now also recognizes inavailable delays for requests of departures that are more than two hours in the future. This allows to less often update data, because data with available delays gets updated more often.
- de_dvb: Added fast stop suggestions by using the URL that's also used by the website's script (returns stops as HTML formatted list).
- de_vrn: Dropped, seems like there are now only departure urls for specific lines.
- fr_gares, dk_rejseplanen, us_septa, de_vvs: Updated to new website layouts (making them pass the unit tests).
- se_resrobot: A new accessor for Sweden with stop suggestions. Recognizes busses, regional/highspeed trains, planes and ships. Can compute delays if the original departure time is given in the remarks section of the result website. But getting a departure document requires to ask the provider for suggestions for the given stop name first, then the "ID" of the best suggestion is used to get the departures. The "ID" is just the index of the stop suggestion, ie. 0 for the best=first suggestion. Getting departures with this service provider can take some more time (~10 seconds).
- international_flightstats: Rewritten with a script, stop suggestions can be retrieved in XML format.
Show an excerpt of the email, instead of just the first lines
We're now stripping all kinds of cruft out, let's see how useful this
turns out for real email usage.
Stripping empty and quoted line's definitely an improvement already,
makes it much more likely that you get to know what you want from
peaking at it.
Activate the category tree
- Add role to get the category path CategoryPathRole
- We can add a category under another category
- We can remove the category and this remove category child and detach
- Fix rename, now we need rename also the path of child category
- Add dnd of category, we can dnd one or severals category in another
- Move Separator from TodoNode to CategoryManager, TodoNode should
accessible only by the todo*model and not by others objects.
Add a second address map provider
Add the OpenStreetMap mapping provider to show addresses of contacts.
Provide a Strigi/Nepomuk-free alternative for IncidenceSearchJob.
This makes invitation handling work even if Strigi/Nepomuk are not available.
With the current implementation it shouldn't be much slower either, since the IncidenceSearchJob is used without a query.
Reconnect the IDLE session if we only lost that one.
In the more likely case of a total connection loss, we do a full reconnect anyway, but if we for some reason only lost the IDLE connection we did not recover until the next full online/offline switch.
make editing connection points more explicit by requiring a click on the shape to activate editing
Committing patch by Thomas Fischer
The patch comes from bugzilla
I will attach a patch for KOffice 2.1.2 to this bug report which improves the "Export to PDF" function in the following aspects:
- If the current document was saved before, suggest the same file name here, too, but replace the file ending with ".pdf". This mimics the behavior of OpenOffice
- The KFileDialogs list of file types is now set via mime types, not a list of extensions
- The dialog's caption is rewritten to match the menu item
o add symetry, mirror and translate mode with GUI
o backport gui from assistant
o reactivate the keyboard shortcuts to change the brush size
o CTRL+SHIFT+click setup the axis center point (same point used for mirror options)
delete active connection shape or connection point when pressing delete or backspace
Make double click on style apply that style
Add rudimentary zeitgeist support
kwlimporter now gets triggered when command line is specified
Add the ability to mark multiple packages at once via the right-click menu.
It still needs a bit of event compression to speed up the marking of multiple packages.
Marking a package is a bit computationally heavy as it is, and without event compression it takes a second or two to mark multiple packages.
Add event compression for marking multiple packages.
For small amounts of packages it doesn't really give a significant increase, but it does make it so that pressing "undo" will undo the last mass-marking instead of undoing each part of the mask marking individually.
Interpret moves with short notation _before_ they are sent to the other protocol.
This enables matches between incompatible engines, for example Crafty vs Sloppy.
Add possibility to notify user of events and offers.
Streaming support for GluonAudio!
The Buffer, Music and SoundReader classes are gone, just use Sound instead.
The Engine class now manages the buffers.
Alure master was imported in the repository as the last release lacks some important functions.
Implement Undo for the FICS protocol.
Support all kinds of algebraic notations for moves
Support from offers from opponent in FICS mode.
AsciiSource: much more inlining
Speedup of this commit:
Delimiters (comment, column) : time
func ptr inlined
No comment, custom: one space : 4.8 3.7
# comment, custom: one space : 5.7 4.2
No comment, whitespace : 6.2 4.0
# comment, whitespace : 6.9 4.2
Introduce major performance speed up
PannerView's content was rendered completely for each scrolling/zooming of the main content.
As the pannerview's content is mostly static, pre-render this and save it into a pixmap.
Move handling of button events config to the same format of suspendsession's one.
This greatly facilitates code sharing and allows to remove quite a lot of redundant code.
Bad news for beta/RC users willing to upgrade though: this might break your profiles' button actions.
With this larger update, we tie down the whole setup process, and waste a lot less cycles resizing.
- zoom method: zooms to specific slice
- less unnecessary resizing (still room for improvement)
- make switching between preview and slice mode work
- fix slice geometry for previews and slice modes
- cache geometries: faster and less dicking around with zoomFactor
- Make applet freely resizable: no fixed aspect mode anymore
- Resizing is also a lot faster now
Make korganizer start faster and use less memory/cpu while running.
There's no need to have two ETMs with every item, one is enough.
before: 22 seconds for 100 pages
after: 8 seconds for 100 pages
with loading the ODF 1.1 odt.
Speed up startup time using Delayed Initialization idiom (see: http://techbase.kde.org/Development/Tutorials/Common_Programming_Mistakes#Delayed_Initialization):
* creating main window first
* all non-crutial initialization procedures delayed through QMetaObject::invokeMethod()
OPTIMIZE: Much of a package's state is guaranteed-static between backend reloads. This means that we can calculate these static flags once on the first time state() is called and store the result for subsequent calls to Package::state()
After Package::state() has been called at least once, subsequent calls will be sped up by 66%. And since we already had an int where we were tracking pin/version override flags, we can use this int to store the static flags and not increase memory usage a bit.
Muon spent half as long inside Package::state() on startup with this optimization. (3% total ->1.5% total time)
Since each row of the PackageList has 3 columns each row has three model indexes.
Create a set of non-duplicate packages from our model index list instead of trying to install packages 3x.
This speeds up change marking on multiple packages by 3x.
cliinterface: Merge create the process creation and execution methods.
Inpired by what Okular does in its comicbook generator, it also makes
sense for us to remove the process we run in the same method (ie
findProgramAndCreateProcess, createProcess and executeProcess have
been merged: the process is now crated, run and destroyed in the same
method, which makes us not need to wory about deleting the process in
While at it, doKill()'s implementation could also be simplified, as we
waitForFinished() right after calling KProcess::start(), so we need
not call it a few more times in doKill().
kdesrc-build has moved to git. Delete everything not essential to kdesrc-build running,
and in addition put in the most annoying exit-program message I can think of to nag people
to cut their ties.
This should avoid confusing scripty while still warning users, without having to leave a
I'd like to thank toma and the rest of the sysadmins for the quick turnaround on the
git.kde.org request, and the i18n team and ReviewBoard maintainer in advance for the
updates to those pieces of KDE infrastructure.
I plan on removing the script from svn entirely around 7 February or so.
Move the kchart and kformula projects out koffice
The kformula and kchart projects are now flake-shapes and
no longer applications and as such they can be developed
much faster in their own git repos.
Version and changelog updated for ver 1.0 release
Let Gluon appear on api.kde.org.
Each namespace have its own Mainpage.dox, so it will have a dedicated page
on a.k.o under the Gluon page which contains documentation only.