This is analogous to commit e053ecdcd5,
which fixes the same problem for the case that the parent folder is
deleted.
BUG: 316335
FIXED-IN: 4.10.2
REVIEW: 109343
Before this change, files were grouped by calendar weeks if they were
modified in the current month and ignoring the "first week day setting",
but by the actual distance measured in multiples of seven days if the
were modified last month.
This not only fixes the "ignores first week day" bug, but also the
problem that the inconsistencies in the algorithm could lead to a broken
grouping in the first days of a month (see screenshot in the review
request).
BUG: 181337
FIXED-IN: 4.11
REVIEW: 108667
This concept is not needed anymore because of the new "determine parents and
expansion levels" approach
Side effect:
Enables treeview for trash, ...
REVIEW: 109191
changes! :)
* Fixed the "Network browser" and "timeline" issues, by using the
KDirLister's itemsAdded(KUrl,KFileItemList) signal -> Use the
given Url to define the parent-child relationship.
* Changed the name of the slot "slotNewItems" to "slotItemsAdded"
for consistency with the signal.
* Use a QHash<KFileItem, ItemData*> instead of a QSet<KFileItem> to
store the filtered data (needed to keep the O(1) lookup for filtered
KFileItems in slotItemsDeleted + needed to fix bug 311912 "After
erasing a filter, some thumbnails randomly disappear")
* Made the determination of the "expandedParentsCount" slightly
simpler - just adding 1 to the parent's level (Also needed to fix the
"Network browser" and "timeline" issues)
FIXED-IN: 4.11.0
REVIEW: 109180
BUG: 304565
BUG: 311912
BUG: 312890
BUG: 315593
There was some code in KStandardItemListView::itemSizeHintUpdateRequired
already that was supposed to prevent an expensive re-layouting of all
items when a preview is received. However, it didn't quite work as
intended because also the "iconOverlays" role changed.
The new approach is to only re-layout if text of a visible role changes,
because this is the only way how the space needed by an item might
change (see KStandardItemListWidgetInformant::itemSizeHint()).
BUG: 315315
FIXED-IN: 4.10.1
REVIEW: 108984
The opacity of the unhovered pixmap must be gradually reduced while
animating otherwise the alpha channel saturates.
REVIEW: 108858
BUG: 299371
FIXED-IN: 4.10.1
The previous approach, which was based on comparing the URLs as
strings, was not only very complex, but also could lead to
inconsistencies in the model, namely, that not all children were
removed from the model when the dir lister reported the parent as
deleted. Later on, this could even lead to a crash.
BUG: 311947
FIXED-IN: 4.11
REVIEW: 108766
1. It seems that it really can happen that KFileItems that we get from
the dir lister's itemsDeleted signal are not in the model any more,
e.g., if a folder where hidden files are shown is left and a folder
where hidden files are not shown is entered. There is no need to
output warnings then.
2. Remove the emptiness-check for the KFileItemList at the beginning.
Even in the unlikely event that we do get an empty list, we return
just a few lines later in the code.
These crashes were caused by the recent commit
ff3267c6dc. It introduced two problems:
a) A logic error in the code that removes the ItemData pointers from
m_itemData that could cause crashes if multiple item ranges are
removed, and there were un-removed items behind the last one.
b) The implicit assumption that any call of removeItems() will actually
result in items being removed in the model. This is incorrect if
the model is first cleared and then the hidden-files setting is
modified, which happens if "Save view properties for each folder" is
enabled, and a folder where hidden files are shown is left. In that
case, the dir lister emits itemsDeleted for the hidden items after
they have been removed from the model due to the folder change.
I'll add a unit test covering these issues soon.
Many thanks to Romário Rios and Hrvoje Senjan for testing!
BUG: 314046
The performance of this method is improved by:
a) Not removing items one by one, but doing it in a way that minimizes
the number of moves to prevent O(N^2) worst-case complexity.
b) Not sorting the removed items using the potentially extremely slow
KFileItemModel::lessThan. We can get the indexes of the removed items
very easily from the hash m_items, and sorting ints is a lot faster.
c) Preventing repeated rehashing of m_items when removing the deleted
URLs by replacing remove() by erase().
REVIEW: 108540
The benchmark executable must be run manually. It is not run
automatically with the other unit tests to prevent waste of CPU cycles
in the not-so-uncommon situation that only test failures attract
attention.
The Nepomuk ResourceWatcher emits 3 signals - propertyChanged,
propertyAdded and propertyRemoved. We should only listen to either the
propertyChanged signal or the propertyAdded + Removed signals. There is
no point in listening to all 3 signals. That will just result in
unnecessary updates.
Additionally, we do not need to listen to the resourceCreated signal.
That is only emitted when we are watching for a specific types, which we
are not.
REVIEW: 108543
In fact, we could use the sorting functions provided by Qt or the STL.
The reason why we have our own is that we want to support parallel
sorting because sorting many items naturally by name can be expensive.
This might make it easier to reuse the parallel sorting code. Moreover,
some the upperBound/lowerBound functions have been removed because
equivalents are provided by the STL.
The intention of the tinting was to make it more obvious in icons view
which icons are selected. However, some icons and previews look quite
ugly with the current tinting value of 1.0 (i.e., the value passed to
KIconEffect::colorize). A slight reduction of this value to 0.8 makes
this a little less ugly. However, the real fix is to remove the tinting
altogether and find something better to indicate which items are
selected.
CCBUG: 309722
The reentrant natural comparison of strings is the only really
expensive operation. Other comparison functions are much cheaper and
might not be reentrant at all. Therefore, we disable parallel sorting
when not sorting by name to prevent crashes and other unpleasant
behaviour.
BUG: 312679
FIXED-IN: 4.10
REVIEW: 108309
When trying to browse a N900 it crashed here because pathA was empty. Which
lead to index = maxIndex beeing -1 and pathA.at(index) crashing.
Reorder the while condition to prevent that courtesy of tsdgeos.
(cherry picked from commit f0c90a47de)
When trying to browse a N900 it crashed here because pathA was empty. Which
lead to index = maxIndex beeing -1 and pathA.at(index) crashing.
Reorder the while condition to prevent that courtesy of tsdgeos.
Fix Bug 308597 - Regression: Renaming a file/folder to something that contains a "/" will result in several message boxes
Uses the same solution as Dolphin-Rename-Dialog does. (KIO::encodeFileName)
BUG: 240820
BUG: 308597
REVIEW: 107681
FIXED-IN: 4.9.5
This reverts 951cb9c35d and
3143acc084, and fixes the crash caused by
nested event loops by delaying the deletion of the KItemListRoleEditor
until the next item is renamed inline.
BUG: 311206
FIXED-IN: 4.9.5
REVIEW: 107606
Copied items should not be shown in gray. Thanks to Christoph Feck for
testing and reporting this regression!
CCBUG: 304615
CCMAIL: christoph@maxiom.de
When renaming inline and starting a drag or invoking the context menu,
a nested event loop will be run. If the role editor loses focus and
emits roleEditingFinished(), we must prevent that deleteLater() is
called because this would delete the role editor inside a nested event
loop which is run from one of its own functions. We would get a crash
when returning from that event loop otherwise.
BUG: 308018
BUG: 309421
FIXED-IN: 4.9.4
Commit ea6a7c09a0 added a workaround for
bug 304986 (high CPU usage because KFileItemModelRolesUpdater requests
previews for the same files over and over again after, e.g., a rename
operation). The KDE/4.10 branch of kdelibs contains the real fix for
this issue (b8f64ca3f4b6311519c21046031d66d9d0a570c6), so the workaround
can be removed.
The real fix is in the KDE/4.10 branch of kdelibs
(commit b8f64ca3f4b6311519c21046031d66d9d0a570c6). We work around the
problem that the KFileItem that a preview has been requested for is
different from the KFileItems in the sets m_pendingVisibleItems and
m_pendingInvisibleItems by refreshing those sets when we detect that
the bug has been triggered.
BUG: 304986
FIXED-IN: 4.9.3
KFileItemModel calls the dir lister's setMainWindow() method to make
sure that the dir lister caches authentication data. However, the method
used to determine the main window (qApp->activeWindow()) is not
guaranteed to yield the DolphinMainWindow or the KonqMainWindow. In
particular, if "Split View" is enabled in Dolphin's settings dialog, the
active window is the dialog, and when it is closed, any later access to
the stored pointer leads to a crash.
A better method is to verify that the model's parent is a QWidget and
then use this widget's window(). I had to make a small modification in
DolphinMainWindow to make sure that it also works correctly when the
view is split (the new view container had been created with a 0 parent
previously).
I tested it in Dolphin and Konqueror and verified that the "main window"
passed to the dir lister is really the application's main window.
BUG: 306459
FIXED-IN: 4.9.3
- Use QSet.erase() instead of QSet.remove() => no expensive Rehashing
- Get rid of += in sortAndResolvePendingRoles()
- Some other small changes
REVIEW: 106577
Added the license header in kitemlistviewaccessible.*
Accessibility headers not included when QT_NO_ACCESSIBILITY defined
Removed unnecessary debug statement
does not expand the dir => Bug discovered: When you drag a
item onto a folder-view-item and then move it away
instantly before the autoactivation event is triggered
(After 750ms), the folder will be opened anyway.
BUG: 305783
REVIEW: 106381
FIXED-IN: 4.9.2
(cherry picked from commit 9ab8bcd6aa)
That commit, which disregarded roles with empty text for the row number
calculation in Icons View, caused the problem that icons might jump
around while information was retrieved asynchronously because previously
empty roles could get a non-empty value, and the corresponding items
would need an additional row in the view.
Thanks to Hrvoje Senjan for testing this feature in master and reporting
this issue early, such that we could fix it quickly and prevent that
other users suffer from this bug!
CCBUG: 304752
Patch 106381 Comment #3:
When "Open folders during drag operations" is enabled, two things happen, both in the DolphinView and in the Folders Panel:
1) When hovering a folder that can be expanded (this is the case for folders with sub-folders in the Folders Panel and in the DolphinView if in Details View mode), toggle its "expanded" state.
2) When hovering a folder that can not be expanded (i.e., a folder without sub-folders or any folder in Icons or Compact View), open this folder in the DolphinView via the KItemListController's itemActivated(int) signal.
The bug described in bug 293200 comment 3 is that 1) is always wanted, but 2) is not wanted for the Folders Panel.
BUG: 293200
FIXED-IN: 4.9.2
does not expand the dir => Bug discovered: When you drag a
item onto a folder-view-item and then move it away
instantly before the autoactivation event is triggered
(After 750ms), the folder will be opened anyway.
BUG: 305783
REVIEW: 106381
According to the backtrace in the bug report, it is possible that
KStandardItemListWidget::paint() is called if the hash m_textInfo has
not been initialised. The widget's index must be -1 in this case, see
KStandardItemListWidget::triggerCacheRefreshing(). It looks like this
can only happen if the item is about to be removed from the view, see
KItemListView::slotItemsRemoved().
I could not reproduce the crash, so I'm not sure why exactly this
happens, but this commit should at least prevent the crash.
BUG: 306167
FIXED-IN: 4.9.2
I'm only backporting the removal of the '-', not the update for the
line number calculation in Icons View, because this is the safest part
of the patch and also the one that fixes the most annoying part of the
bug.
Thanks to Todd Jennings for the patch!
BUG: 304752
REVIEW: 106304
(cherry picked from commit 20b0cb68bf)
The crash was caused by a null pointer dereference when, e.g.,
minimizing Dolphin. The root cause was that
KStandardItemListWidget::closeRoleEditor() was called twice: once when
the role editor loses focus, and once again when the window is resized.
After m_roleEditor was set to 0, the second call dereferenced this null
pointer. I think the best solution is to disconnect from the role
editor's signals when the editor is not needed any more by the
KStandardItemListWidget.
CCBUG: 304524
(cherry picked from commit a9c2bdc3b5)
The crash was caused by a null pointer dereference when, e.g.,
minimizing Dolphin. The root cause was that
KStandardItemListWidget::closeRoleEditor() was called twice: once when
the role editor loses focus, and once again when the window is resized.
After m_roleEditor was set to 0, the second call dereferenced this null
pointer. I think the best solution is to disconnect from the role
editor's signals when the editor is not needed any more by the
KStandardItemListWidget.
BUG: 304524
FIXED-IN: 4.9.1
For top-level roles like "Name" and "Size", the group translation is 0.
In that case, the "group" member of the corresponding RoleInfo struct
must be an empty string. This is expected by the code which generates
Dolphin's menus, which group the top-level roles into a sub menu with
the name "(I18N_EMPTY_MESSAGE)" otherwise. This is a recent regression
caused by 199fabbaa8.
Thanks to Christoph Feck for investigating this issue!
CCBUG: 305228
(cherry picked from commit 168d0511c7)
For top-level roles like "Name" and "Size", the group translation is 0.
In that case, the "group" member of the corresponding RoleInfo struct
must be an empty string. This is expected by the code which generates
Dolphin's menus, which group the top-level roles into a sub menu with
the name "(I18N_EMPTY_MESSAGE)" otherwise. This is a recent regression
caused by 199fabbaa8.
Thanks to Christoph Feck for investigating this issue!
CCBUG: 305228
Remove comparison of QAccessibleInterfaces - these are pointers that get
instantiated again and again, so they cannot be compared. Use the object()
function in QAI instead for comparison.
The base class implementations are actually good and without some problems.
Save one pointer's worth of memory by not having m_container.
rect was returning a relative rectangle always.
State was wrongly returning HasInvokeExtension which can lead to crashes.
Moreover, make sure that the selection rectangle is large enough to
include the stars.
Patch by Emmanuel Pescosta. Thanks for the patch!
CCBUG: 302624
REVIEW: 105830
(cherry picked from commit 146c77eaa2)
Moreover, make sure that the selection rectangle is large enough to
include the stars.
Patch by Emmanuel Pescosta. Thanks for the patch!
BUG: 302624
FIXED-IN: 4.9.1
REVIEW: 105830
Use QStyle icon sizes instead of hard coding 16 or 32 pixels.
Using styles that follow KDE settings (every KStyle, e.g.
Oxygen) the icon sizes can be configured by the user in
System Settings.
The size for the icon/details view can be later changed
by the user, overriding the defaults. The icon size for the
places panel is not (yet?) configurable separately.
(Not resolving the bug, see comment in review request)
CCBUG: 301959
REVIEW: 105548
(cherry picked from commit 79211f6e48)
Use QStyle icon sizes instead of hard coding 16 or 32 pixels.
Using styles that follow KDE settings (every KStyle, e.g.
Oxygen) the icon sizes can be configured by the user in
System Settings.
The size for the icon/details view can be later changed
by the user, overriding the defaults. The icon size for the
places panel is not (yet?) configurable separately.
(Not resolving the bug, see comment in review request)
CCBUG: 301959
REVIEW: 105548
This resolves a regression caused by the Places Panel rewrite. There is
a small glitch left when reordering items (dragging below the last or
above the first item only shows the drop indicator when first dragging
out of the item and then back), but I prefer not to fix this glitch
right now because this would require a more intrusive change, and I do
not want to risk regressions because is not much time left to fix them
before 4.9.0 is released.
Thanks to Peter Penz for providing some advice about this issue.
BUG: 302557
FIXED-IN: 4.9.0
(cherry picked from commit f4c9600251)
This resolves a regression caused by the Places Panel rewrite. There is
a small glitch left when reordering items (dragging below the last or
above the first item only shows the drop indicator when first dragging
out of the item and then back), but I prefer not to fix this glitch
right now because this would require a more intrusive change, and I do
not want to risk regressions because is not much time left to fix them
before 4.9.0 is released.
Thanks to Peter Penz for providing some advice about this issue.
CCBUG: 302557
If a file with an unknown extension gets renamed, it is assumed
that the letters after the last point represent the extension and
won't be selected.
BUG: 301893
FIXED-IN: 4.9.0
The regression has been introduced when hiding the DolphinDirLister
inside KFileItemModel. Now the signal urlIsFileError() gets forwarded
to the container again where the file will be opened.
BUG: 301757
FIXED-IN: 4.9.0
- Don't use the selection-tinted icon when dragging a single item
- Decrease the vertical gap for the hotspot
- Prevent drawing tree-lines when dragging items for the Places Panel
CCBUG: 301624
If the "isExpandable"- or "size"-role is shown for a folder, the
number of sub-items must be watched to be able to update the
state of those 2 roles.
This fixes the issue that e.g. no expanding-toggle is shown
if an sub-directory has been created inside an empty directory.
BUG: 293972
FIXED-IN: 4.9.0
... otherwise no expansion toggle will be shown. The implementation
is a compromise between performance and correctness: If the directory
contains links that are not directories, an unnecessary expansion
toggle will be shown. But this is better than showing no expansion
toggle in case if the directory contains links that point to other
directories.
BUG: 299997
FIXED-IN: 4.9.0
When emitting the roleEditingCanceled-signal it must be prevented
that a roleEditingFinished-signal is emitted because of some
action taken in the receiver-slot.
BUG: 301480
FIXED-IN: 4.9.0
Fix issue that cancelling the inline-renaming by clicking
on the other split view applies the wrong focus.
Thanks to Frank Reininghaus for the initial patch and noticing
this issue.
CCBUG: 301252
- Don't use an outdated KFileItem when renaming an item more than once
- Use the same font as the view
- Don't lose the focus when an editor-popup is shown
BUG: 299327
BUG: 298883
BUG: 301253
FIXED-IN: 4.9.0
If the current folder is collapsed, pressing the left arrow key now
moves the focus to the parent folder.
CCBUG: 300582
(cherry picked from commit 93daadae28)
The problem was the following: While a QMutableSetIterator iterates the
items in the QSet m_filteredItems, all items that match the new filter
are removed from the set. However, to ensure that the iterator still
reaches all items in the set, one must not use QSet::remove(), but
rather the iterator's remove() method (see QMutableSetIterator docs).
CCBUG: 300504
(cherry picked from commit 288473a96c)
The dropping has not been implemented yet, however in the context of
this step the creating of the drag-pixmap is now forwarded to
the item-widgets. This allows creating some optimized dragging-pixmaps
e.g. for the details-view, where only the name and icon should
be provided as drag-pixmap.
If multiple ranges are inserted or removed, the updating of
the group-headers may only be done after _all_ ranges have been
inserted/removed and not after each individual range. Otherwise
the layouter-cache is not in a consistent state yet.
Follow the suggestion from Vishesh Handa and copy the ResourceWatcher
code from kde-runtime instead of creating a dependency. This will of
course get removed again as soon as Nepomuk's resource watcher will
get a public API.
Related changes:
- Animate changed items for the details-view in case it is not expandable
- Remove the hardcoded "isHidden"-code in KStandardItemListWidget and
allow derived widgets to define themselves what means "hidden" within
their context.
The current code needs a lot of bugfixing, but lets make this in smaller
steps during the next days...
Implement adding, editing and removing of entries. Note that the
result currently is still not stored in bookmarks.xml (this needs
to wait until the hiding is implemented in the model).
When changing the sort-column the sort order should always be reset
to 'ascending' to by consistent with the QListView behavior.
BUG: 298830
FIXED-IN: 4.9.0
This fixes two problems:
1. KItemListKeyboardSearchManger can cancel the current search when a
new folder is opened (note that this action removes the current item
from the view).
2. The view can underline the new current item (which is the item that
used to be below the removed item). Note that this did not work
before because the view did not receive a currentChanged() signal in
this case and therefore did not update the "current item" status of
the new current item.
CCBUG: 297488
CCBUG: 298782
REVIEW: 104709
(cherry picked from commit 68ce395a19)
This patch restores the Dolphin 1.x behaviour that Esc clears the
selection (Dolphin 2.0 did not have an easy way to clear the selection
using the keyboard so far). Moreover, Esc now also cancels the current
keyboard search. This fixes the problem that the only way to cancel the
search is to wait until the timeout has expired.
CCBUG: 298742
CCBUG: 297458
(cherry picked from commit d3b1312d3e)
It turned out that the longer timeout, introduced recently in
02eab49b2d, not only made multi-letter
keyboard searches easier, but also had some unwanted side effects. I
hope that 1 second, which is between the previous value of 5 seconds and
the pre-KDE 4.8.2 value of 0.4 seconds, is a compromise which will fit
most users' needs.
We will try to improve the situation further in future releases by
providing visual feedback about the keyboard search, but such a change
would not be suitable for a bug-fix release.
(cherry picked from commit d8732a59d3)
Up to now the view-engine only provided a model-implementation that
supports file-items. The view-engine always had been designed to be able
to work with any kind of model, so now a KStandardItemModel is available.
The plan is to convert the places panel to the new view-engine. It should
be no problem to fix this until the feature freeze - in the worst case
the places-panel code could be reverted while still keeping the
KStandardItemModel changes.
The attached patch sets the main window on the main directory
lister in DolphinView and KIO jobs in DolphinMainWindow to ensure
that login data for remote protocols such as sftp, ftp are cached
properly for the duration of the application. Otherwise, the end
user is going to end up being unnecessarily re-prompted to enter
password login information.
Thanks to Dawit Alemayehu for the patch!
REVIEW: 104614
FIXED-IN: 4.9.0
Showing the whole filename unclipped seems to be a good default,
however for users with a lot of files that have extremely long names
this might get a problem especially in the icons-view.
- Allow to limit the maximum number of lines in the icons-view
- Allow to specify a maximum width in the compact-view
(No limit is required for the details-view, as the name is shortened
automatically to show other columns)
BUG: 288596
FIXED-IN: 4.9.0
- Move all private headers from the kitemviews-directory into
the 'private' subdirectory.
- Get rid of DolphinDirLister and just use a directory-lister
internally in KFileItemModel.
- Minor interface-cleanups for signals
If the sorting is done for data which is resolved asynchronously
(e.g. rating), it is important to give a visual feedback about
the state of the sorting. This is done now by a progress
indication in the statusbar.
Also optimizations for "Sort by type" have been done: Although
resolving a type can be expensive in the most often case it is a
very cheap operation. So it the sorting is done by type, try
to resolve the type synchronously for at least 200 ms to prevent
a asynchronous resorting. This is usually sufficient to have
resolved types even for directories with several thousands of
items.
BUG: 292733
FIXED-IN: 4.9.0
The feedback on bugs.kde.org has shown that the previous behavior
(= show size with best-matching unit) is preferred by most users.
I initially wanted to make this configurable, but for implementing
it in a non-hacky way extending KLocale from kdelibs would have
been required. I'm not sure whether the usecase in Dolphin justifies
having such a configuration in KLocale - however as kdelibs is frozen
at the moment this is no option and the old behavior has been
restored.
BUG: 289850
FIXED-IN: 4.9.0
It is up to the plugins to decide what sensible size is required for
creating a preview, this cannot be decided by the user for all
plugins. A maximum size only makes sense for remote files, where
the user implicitely decides how much of the bandwith may be used
for having previews.
BUG: 273226
FIXED-IN: 4.9.0
