|Commits||1868 by 165 developers|
|Bugs Opened||219 in the last 7 days|
|Bugs Closed||234 in the last 7 days|
|Hugo Pereira Da Costa||
|Albert Astals Cid||
Internationalization (i18n) Status
|Chinese Traditional (zh_TW)||
|Hugo Pereira Da Costa||
|philippe roubach free fr||
|Albert Astals Cid||
There are 50 selections this week
Fix deadlock while parsing projects with deep include trees.
This complicates the code quite a bit, but I could not come up
with anything simpler.
We create a sorted list of includes by depth and traverse that
in a depth-first manner. Furthermore we remember what files are
imported in which files and ensure that these imports are available
when parsing a file as otherwise include guards and cyclic references
would break or lead to infinite recursions.
For every file or import we set the UrlParseLock and check whether
the DUChain information is up to date. If not, we traverse the whole
translation unit and look for any declarations that belong to the
This ensures we never nest the UrlParseLocks and thus should not
Fix: Do not crash on functions with call count 0.
Cherry-picked from master.
Call count 0 may exist with Callgrind when measuring starts after
the function was called but when the function is still active,
ie. not returned yet. For "cost per call", it makes sense to show
the full cost while the function was active.
Optimize: Don't put DUContext::Other contexts into the symbol table.
This is inherited by the declarations in these contexts which is
good as we don't want to pollute the costly global symbol table with
Show locations for all breakpoints even if they have neither
file nor line information.
Otherwise we'll be showing valid breakpoints with empty locations. Also
these breakpoints now'll be automatically inserted on debugger's restarts.
Don't insert the same breakpoint multiple times while starting. Part 2.
First compare "fullname" and "line" entries and only if it fails fall
back to "original-location", as the latter can contaion literally
anything. Note that even a better solution'd be to store a full path
in the first place (if available).
Do not resetSearch if we're closing down
It's useless as it will be reset anyway on the close down procedure
and it will cause a crash if you are unlucky.
* In the document observers qset pageview comes before
* We are closing and notifySetup with an empty vector
* pageview empties its vector in notifySetip
* part sees the notifySetup and resets the search
* resetting the search calls another notifySetup with the non empty page vector
that will be deleted asap when we go back to the first notifySetup
There are another billion ways to fix this:
* Don't allow nested notifySetup
* Clear the running searches before sending the first notifySetup
* Empty the pagesVector before sending the first notifySetup
But i feel this is the less invasive one (even if it may not be the most technically correct)
Fix preview size for some backends
Some backends don't return data until everything is scanned.
handle update file timestamp option from Setup/Metadata when file are changed outside metadata, as transformation through icon-view (JPEG files) or when
files are overwritten from Image Editor. BQM already manage this option properly. Import tool is not affected by this settings (as import or export file always preserve file timestamps)
Toogle on option from Setup/Metadata config pannel to on, to prevent all dysfunctions with extra application which use file timestamps to detect automatically file modifications.
Make KFileItemActions the parent of the actions it creates
This prevents the problem that the destructor of KFileItemActions may
delete dangling pointers if the widget that is set with setParentWidget
is destroyed before the KFileItemActions instance (note that this widget
was the parent of the actions before this commit, but that
KFileItemActions stll remembered them in d->m_ownActions and deleted
them in the destructor.)
make killline work with current selection too
Kill any running preview jobs before starting a new one
If loading a preview takes long, it was possible before this commit
that a preview for a new item was requested before the first preview
was shown. In that case, there was a race condition, and the first
preview to arrive stayed in the Information Panel.
This commit fixes this by keeping a pointer to the preview job and
killing it before starting a new one.
Workaround issue in libotr 4.0.0.
This still has an issue with fragmentation as long as libotr 4.0.1 is
not released, but at least it doesn't crash.
Make the height of the Preset Chooser Popup 600px
This value is consistent with the height of the PaintOp configuration
Thanks David Revoy for suggestion!
Fix the outline size in hover mode
In hover mode the pressure must be emulated to be 1.0, otherwise
we will not see anything.
Probably fix the Ascension sensor in Hover mode
This patch passes the real tablet event to the outline generator
in hovering mode, so if your tablet supports rotational sensors it
the outline will be generated accordingly.
Please someone test it with a proper Tablet Device.
Fix Color Smudge op in Dulling mode
We must not store reading random accessors when someone else in writing to
the same device, because the cache of the random accessor will become
Fix calculation of the shape's insets when it has Miter joining
The insets should be not smaller than the limit of the Miter joining
Fix the initialization of the default bounds of the Paint Layer
I don't know why the line was commented out :(
Don't show an error when autosave fails in Krita
We'll retry it later again, since we cannot save if the image is locked,
so don't worry the user with alarming messages.
On doing a full prescale, clear the projection image first.
Disregard opacity with fg/bg fill
Bucket fill still uses the currently set opacity, but the shortcuts to
fill the entire layer with background or foreground colors should not
use the global opacity.
Fixing playing of files from command line or by any other means.
PMC was failing to play the songs which were passed from the command line.
It was playing the songs from the current playlist. With this patch it
rectifies the issue.
Signed-off-by: Sujith Haridasan
Check both pid and gtk app name to identify XUL applications
Also check for a non zero length text here.
The QChar* pointer could be not null but pointing to the single char '\0'
due to how DOMStringImpl is implemented.
(Forward-port kdelibs commit e71c6b7b97c80afa284b854734b0a7d35868e093)
kio_sftp: Support ECDSA keys.
This migrates to the API of libssh 0.6.0.
Added a nepomukcleaner job for removing duplicate series.
Add an optional Baloo migrator
FileMetaDataWidget: Hookup the tags, ratings and comments
Add xterm's bracketed paste mode
Allow xterm's bracketed paste mode to work
When enabled by '\e[?2004h', pasted data is surrounded by \e[200~ and
\e[201~. This is mainly for text editors to temporarily switch off
autoindent and line wrapping.
Allow to undo/redo archive mail
improvements to the attachment editor.
better ability to switch between inline and linked attachments
first draft SFTP implementation
Add support for nonzero fill rule for experiment brush
Thanks David Gowers for preparing this patch!
Add a margin to KisGradientSlider
So the full triangles are shown at the edges
Added two triangular cursors for the painting
(it also adds two outline-enabled counterparts)
Optimize: Skip function bodies when looking for visible decls only.
This decreases the parse time of my test project by ca. 25% from
~13s to ~10s. It is only applicable to VisibleDeclarationsAndContexts
or less features, but that is the default for project files anyways.
Refactor slotScheduleIndexing to make it clearer when each queue runs.
Along the way, fix a couple of odd edge cases:
1. The file queue would sometimes not be started correctly
2. The cleaner would stop when going onto battery, but if we
were already on battery when starting, it would still be started.
Change this so the cleaner runs even when on battery.
Also make some adjustments to the index cleaner so that it works when
resumed: these are only necessary because it is not yet implemented.
Make the specific color selector more responsive
KDM goes the way of the Dodo.
It's exactly, to the day, 6 years ago that we released "KDE 4.0", while
this is of course entirely unrelated to this commit, let's celebrate
this anniversary with the deletion of kdm from kde-workspace.
Happy anniversary from Barcelona!
The Plasma Team
Move WidgetExplorer to be a private QML plugin
Instead of registering WidgetExplorer inside the shell, create a new qml
plugin in the org.kde.plasma.private namespace.
This is useful for running widgetexplorer in qmlscene.
This makes testing, debugging and profiling considerably easier.
Screenlocker: don't set the mouse cursor when grabbing the mouse
Setting the cursor to ArrowCursor when calling XGrabPointer() prevents
the Screen savers from blanking the mouse cursor.
With this change they can blank the mouse cursor again.
Phone number shown next to contact name for calls/sms
Add a new Domain module containing the Artifact class
This marks the start of hopefully the right redesign of Zanshin. It will
be the last one, I swear.
- Reinstate code from "Add VAT split to file import", inadvertently removed by this change.