28th November 2010by Danny Allen
Alexander van Loon
|Commits||2338 by 196 developers|
|Bugs Opened||596 in the last 7 days|
|Bugs Closed||770 in the last 7 days|
|Nuno Fernades Pinheiro||
|Iñigo Salvador Azurmendi||
|Jonathan Michael Thomas||
Internationalization (i18n) Status
|British English (en_GB)||
|Low Saxon (nds)||
There are 182 selections this week
Always shutdown monav-routingdaemon on shutdown on Maemo, even if we didn't start it in this instance.
This makes it more unlikely that USB mounting is being blocked, or at least provides the workaround of starting and terminating Marble once after a crash to get rid of the block.
- Make it visually clear that the new option belongs to tied zoom
- Improve the wording from "Tie Between Tabs" to "Tie Across All Tabs
(though as a non-native speaker I feel a bit unsure changing something defined by a native speaker => please undo the change if you don't agree)
Improve rendering of items for projections which repeat in the x-direction: Items are now drawn at geo-coordinate positions and not at pixel positions.
This way, Marble renders them in all places on the screen which represent these coordinates.
However, the current check for whether items are visible does not seem to work in all cases and needs to be investigated later.
Forgot that some language have more than one localisation (en_us).
Splitting the string and getting the first value should do the trick.
Thank missive for testing.
Switch to new Facebook API for requesting application permissions.
This have prevented all new users since July 2010 to upload photos to specific album or download photos (own or from their friends).
Conditionally changes KFilePlaceEditDialog's title to "Add"/"Edit" to reflect the operation that is actually being performed.
In the item view constructors, call setFocus() only after m_extensionsFactory is initialised.
This fixes a possible crash in Konqueror when changing the settings or manipulating tabs.
Remove selection only when receiving input
Port away from kde3support (k3command/k3commandhistory -> qundocommand/kundostack)
Fix crash due to "delete cur_item ; cur_item = foo" without telling the caller of the method
who passed cur_item by pointer and kept using it afterwards. Interestingly this has been in the
Load parts using RTLD_GLOBAL so that the domtreeviewer can catch exceptions from khtmlpart.
Otherwise it just crashes.
Backport of two important fixes:
* Fixed KInotify which sometimes thought / was deleted which resulted in removal of all indexed files from the index.
* Do not delete all graphs which contain information about removed files. Instead remove the resources since the
graphs might also contain information about other resources.
backporting r1197040 "clipping error on VD change"
backporting r1183245 "secure clientGroup pointer accesses"
Implement the "past names map" bug-compat feature from HTML5;
needed by libero.it e-mail app (the old one)
Do not free unallocated data in case of corruption
Daniel, if you manage to find the XCF (GIMP format) file that
causes the crash and it loads fine in GIMP, please attach it.
It is possible that there is some weird format that KDE does
not handle yet.
Temp fix for some crashes when changing profiles in Konsole Part.
Fix two bugs that were compensating each other in the unittest here:
1) kbuildsycoca would not emit databaseChanged("...") for resources where desktop files had been deleted
2) a name clash ("screensaver.desktop" in both servicetypes and services) made kbuildsycoca4 always
And of course the unittest was for "deleting a desktop file in the services resource" so it was passing by pure chance (and not on Volker's machine which didn't have both desktop files).
Fixed 2) by using resource+path as the key in that dict rather than just path.
and fixed 1) by looking at g_ctimeDict at the end, to see which resources are still there
(which needed the fix for 2 in the first place)
The fix for 1) uncovers a new bug though: databaseChanged("apps") is always emitted, so we never go into the fast path of "no need to save to disk"; I'll look at this tomorrow.
More fixes to locale kcm, and now re-enable saving!
Better solution for checking if config changed, but would be even easier if KConfig/KConfigGroup provided an isDirty() call (and a way to open as read only would be nice too).
Languages fixed to load/save correctly, but I still can't get the KActionSelector widget to work properly, to save tearing my hair out anymore I'm going to tear it out instead and do it myself.
Ensure that KStandardDirs always returns resolved paths (when $HOME or $KDEHOME is a symlink),
even when returning paths that don't exist yet. Otherwise, via caches (the kstandarddirs one
or in applications), we can end up with a different path once it does exist, and comparisons
will fail (e.g. the one in KConfigPrivate::parseConfigFiles which detects the local file).
(So this fixes "kconfigtest testRevertAllEntries" when $HOME is a symlink, but has its own
setup-independent unittest now)
Don't crash when someone (e.g. kmail) passes in some html w/o a KHTMLPart that contains a named iframe.
It'd be best if HTMLQuoteColorer in kmail 4.4 were changed to use the part just in case, though.
Allow changing to pages that have no module proxy
KSettings::Dialog adds items without module proxy, so we cannot change from or to those.
The check can be removed completely, because "resolveChanges" already makes sure it does only ask for changes on pages having a proxy.
Patch from George Metaxas, Thanks!
This patch fixes bug 74938, which is related to the way in which the
systray icon of KMail behaves. Currently, clicking on the systray icon
has the effect of minimizing the KMail window, irrespective of its
position in the window stack, and maximizing it if it was previously
minimized. In the latter case, the current virtual desktop is changed to
the virtual desktop where the KMail window was minimized. As noted in
the corresponding bug entry, this behavior is different than the
corresponding behavior of other KDE applications (e.g. Akregator).
The patch modifies the behavior of the KMail systray clicking handling,
by making it identical to other applications (e.g. Akregator). When
the KMail window is the top level window in the virtual desktop, it
is minimized. If it is visible in the desktop, but not at the top,
it is brought to the top. If the window is minimized, then it is
maximized to the current virtual desktop.
Fixed a full refresh for Clone layers
Now the Async Merger knows about one more type of layer position:
N_EXTRA. Such layer will be counted in a need rect chain and will be
updated (original+projection), but it will not be merged into the
final projection of the image. This is done for preliminary updates of
the sources of Clone layers.
Btw, i would suggest everyone to perform 'make clean' in
./krita/image/ folder, because many parts of walkers are
inlined. There is no need to do a full clean, just in ./image/
Fix the bug and implement the KMyMoney header functionality in a much nicer way.
Instead of hiding the standard header and having the KMyMoney header in all the views remove the standard header from the layout and replace it with one instance of the KMyMoney custom header (the text is adapted after a page change).
Improve the layout of the budgets and forecast views for a better vertical space usage.
The only thing that was preventing the usage of KMyMoney on 1024x600 was the settings dialog which is fixed by this commit.
For the status and tool bar reported for this bug they can be shown if the KMyMoney header is hidden using the settings.
So a user with 1024x600 can hide one of the following 3 items to make KMyMoney fit into 600 pixels in height:
- the title widget
- the tool bar
- the status bar
Fix Cost breakdown: Planned startup/shutdown cost is included in actual cost.
* call adjustSize() direct in postLayoutHook() to be sure we first adjust the size and then position the shape (where it's size was used).
* cache the documentSize. An even better way would be to calc the document's size while layouting but that's for later.
* on centered vertical-alignment proper position in the middle. This fixes page #21 in the a.odp.
Task performance calculations:
* Add startup and shutdown cost to bcws/bcwp/acwp calculations,
part of Bug 257413 - Different material costs inclusion in BCWS of kplato.
* Don't add a day too much when calculating a map of effort/cost pr day.
* Handle non-specified end time when fetching actual effort/cost.
factory() can be NULL according to View.cpp (grep for factory()).
This should fix a reported Rekonq crash while browsing a ods file.
Fix the bug plus as an extra feature to display the institution value with the negative color if it's value is negative.
Show entries in normal color (again) if their balance becomes positive while another account is added.
Backported to stable branch
Fix reading of WavPack streams without a length information in the header
When the WavPack's total_samples header fiels contains -1, try to find the final block and get the number of samples from there as block_index + block_samples.
If a download did not report a totalsize do not use the finished chunks to calculated its processed size, as none has been set to finished.
Several fixes for queue handling.
Also remove the confirmation dialog, since the confirmation really needs to occur before the transaction is added to the transaction stack, and I don't have a sufficiently accurate way to predict download size without actually marking things for installation
Rework clirarplugin's line reading logic again.
The code looks more like a state machine, which is good and should make it
easier to understand.
This also made it possible to add some code to check the column names written
before the header lines and decide whether we're using unrar-free or unrar
based on that (which should fix Ark crashing when being used with unrar-free).
If we're using unrar-free, we never try to parse the third file name entry
line, as unrar-free never outputs it.
It should be hopefully safe to backport, but I'll wait at least a few days. As
I'm unsure whether it's OK to backport it, the FIXED-IN version will be 4.6.0.
- Menu option to specify if tied zoom spans tabs.
- The tied zoom button *toggles* tied stat in current tab. Since individual plots can be tied and untied, it is not a latching button.
- Add a head to curves, so the most current data can be identified in XY plots that do not have monotinic X vectors.
Adding external links to Webelements and the Jeferson Lab pse.
Unleash non-destructive editing:
The interface is now fully adapted to versioning (includes
a first toolbar action cleanup). Yes, Save/Save As is gone.
Icons perfectly fitting to our concepts are missing, so I'm using
what's available. (I filed a request for icons, but dont know when/if it will be done)
I plan to extend the setup a bit, possibly allowing to switch off
non-destructive completely for those that really want,
and to define the default close action.
(We had planned to default to "save current" when closing the window,
but now I think that per default, the user must be asked)
You'll also see that "Revert to original" kind of changed its nature.
There's a "Load Original", but that's like opening a file.
Then there's a "discard changes", which is effectively like clicking
Undo to the end. Something like "remove the current version, delete
the version file and all its intermediates up to the original" has a lot of
implications, if other files were forked in between.
Make deleting emails work, add mark as spam & todo
Two things are interesting in this commit:
1) We use the deleted flag for deleting emails, let's see how clients
handle this. I *think* this solution is quite elegant, but we might want
to move items into the trashfolder (which?!) instead. SpecialCollection
might help here. Or maybe we want to _really_ delete items?
2) The context menu offers some additional options to mark emails, spam
or task. These options would clutter the UI too much, but are nice to
have available for power users (those are more likely to find them there)
Report supported file patterns on WinCE, where we do not have sycoca.
Supports KDE3 way of starting man-pages (#term) and info-pages (##term).
Remove empty autosave files as well.
* implement ODF draw:auto-grow-width and draw:auto-grow-height.
* turn changing resizeMethod into a QUndoCommand to be able to undo/redo
* this makes those a.odp working if it got loaded+saved once by ooimpress what is the case cause ooimpress has a bug where it always sets auto-grow-height=true
Big overhaul of progress reporting for transactions:
- Add a queue so that you can browse/install other packages while downloading
- Send out custom signals from the ApplicationBackend
This allows us to report progress in both the application delegate extender and the application details view, without progress being reported in all delegates or extenders.
Removing nepomukshell from kdereview. Moved to extragear.
Nepomukshell can be found https://projects.kde.org/projects/extragear/utils/nepomukshell
Google Code-in task solved by Cezar Mocan, consisting of code refactoring in the ObservingList class.
He removed significant parts of duplicated code and also identified a bug in the trunk (the context menu
was not showing for object on the session list) and fixed it pretty fast, even though he was not required to do so.
Results: cleaner and efficient code, one bug solved.
Replaced the simple query-based token detection with the one written by Mikhail Kotelnikov which uses a token tree of all named entities.
This is faster and also handles tokens which contain strings.
But on the other hand it does not handle substring matching yet and uses an in-memory tree of all named entities.
Redo the calculation of the bounding box in a better way.
The new way means that the bounding box is only recalculated when
there is a drawing operation. This means that changes to the
window/viewport origin or extension with no drawing are
disregarded. It also means that there is no recalculation when both
org and ext are changed in between but just after both have changed.
This probably means that we get a much better bounding box in many
QtCore4.pm, KDECore4.pm: Allow client modules to tell QtCore4.pm about an array type they implement, for ambiguous method resolution.
handlers.cpp: Fix the from Perl marshaller for classes to use a version of Smoke::cast that correctly handles module boundaries.
util.cpp: Signals can be overloaded, but we want to check to make sure that the number of arguments passed to the signal are the same the number the signal accepts. So if we find a signal with the correct name but the wrong number of arguments, keep looking for another one.
util.cpp: Some weird cases when using the kperlpluginfactory ends up with entries in the pointer map that have a refcount of 0. Running hv_delete causes Perl to abort. So check the refcount before trying to delete them.
kperlpluginfactory.cpp: Use a "use lib" line instead of passing "-I" in argv to perl_parse. This way we can support loading in multiple plugins.
kdecore4handlers.cpp, plasma4handlers.cpp: Add a bunch of list handlers, copied from Ruby.
Update kdeui examples to not use the Qt4:: namespace anymore.
Update the analogclock plasmoid example. Now complete with calendar!
Here we go again... :-)
Adapt digiKam coding style guides:
This is a huge patch to finally have a common coding style in the source code.
This patch adds:
- curly brackets to all single line statements
- left pointer / reference operator style (char* bla)
- indent 4 spaces
- convert tab -> spaces
I did not change the libs from 3rdparty and the ones from libs/jpegutils.
Code compiles fine under Linux, there shouldn't be code breakage because we
only reformat the code and add missing brackets. If code is broken now
(logic errors due to missing brackets), it was broken before.
I add this as one single patch so we can easily revert it, if needed.
I think it is really important to have one common style in the code, and that we don't mix it.
I will also add some comments to the HACKING file later on.
can you check if all compiles fine under Windows and Mac?
- Replaced the direct use of webshortcut plugin configuration files with the new APIs in KUriFilter.
- Changed the code that launches the webshortcut configuration dialog to use KRun::runCommand.
- Connect to dbus signals to be informed of webshortcut configuration changes.
- Commented out all the debug statements. They can be uncommented when needed.