| Age | Commit message (Collapse) | Author |
|---|
|
Prior to this commit keyboard controls and behaviour of Dolphin's main
view were identical no matter if selection mode was enabled or not.
While selection mode makes it impossible to accidentally clear the
selection by singular mouse clicks, any press of an arrow key on the
keyboard would still clear the full selection which goes against
selection mode's objective.
Furthermore, keyboard-only users had no reason to ever enable selection
mode because it made no difference to them.
This commit changes this by offering a changed control scheme for key
presses while in selection mode. Arrow key presses without modifier now
only move focus between items but do no longer clear or change the
selection. Similarly, Page Up/Down, Home, and End key presses only move
keyboard focus. Enter, Return, and Space key presses now only toggle
the selection for the current item.
The above controls are however mostly unchanged when combining them
with Modifier keys like Shift or Control.
The type-ahead feature is also changed in selection mode to only move
keyboard focus without changing the selection.
This way keyboard users are less likely to clear their selection by
mistake. Regression tests are added for these selection mode controls.
The code changes to change this keyboard behaviour are quite minimal.
Most of the added code is for making selection mode accessible. That's
because we need to make sure the changed control scheme is properly
announced and communicated or a blind user will be left utterly
confused why the normal keyboard controls "stopped working".
Enabling or disabling selection mode is announced to accessibility
software. Furthermore whenever focus goes to the main view, the
selection mode state is also mentioned when active.
BUG: 458091
|
|
If filename of an item was updated previously, it would modify the model
before the file was actually changed. This led to the model calling
a signal that would try to run a previewjob, but since the filename
is not actually changed yet on disk, it would fail.
This patch moves the model updating after copyjob. Copyjob
will take care of the file renaming if there is already existing file.
We just need to update the model correctly after the job has succeeded.
BUG:497555
|
|
If viewPropertiesString is empty, return a nullptr.
This will later used in the stack by the defaultProperties call.
In defaultProperties, if we can't find the global directory,
create new one with a tempfile. If tempfiles can't be created,
use default instead.
This will ensure that view settings are saved and loaded correctly
if user has separate view properties per folder.
This will also add an unit test, where we create a global directory,
modify it and make sure the changes are reflected in the unmodified
folder.
BUG:495878
|
|
Suppose we are renaming file i and the new name will be sorted after
file i+1. We are now pressing ArrowDown to immediately start renaming
file i+1. However, because of the sorting we would not actually end up
renaming what used to be file i+1. What would happen is that editing
would be started in dolphinview.cpp:2065. However, after 100 ms the
timer in KFileItemModel would fire, resulting in the model emitting
itemsMoved(). This would trigger doLayout() in
KItemListView::slotItemsMoved(). doLayout() resizes the
KItemListWidgets, wich causes the renaming to be canceled in
KStandardItemListWidget::resizeEvent().
Now, we start a new renaming operation for the correct widget after the
relayouting is complete.
|
|
This one-liner makes sure keyboard focus always moves to the active
view when a place in the places panel is activated. Previously
focus would remain on the places panel if the activated location
was identical to the already displayed location.
This added consistency is helpful because keyboard-only users will
get used to be able to move from the places panel to the view by
pressing the Enter key, even though this does not always work
prior to this commit.
This issue was identified in an accessibility scan done by HAN
University of Applied Science.
|
|
Currently natural sort compares the entire filenames
(basename.extension) when sorting. This causes eg.
"a 2.txt" to appear before "a.txt" when sorted by ascending.
This is unintuitive since people prioritize basenames more
than file extensions.
Instead, change natural sort to compare by basename only and
fallback to comparing extensions if basenames were equal.
This change causes "a.txt" to appear before "a 2.txt" and
matches how other platforms such as GNOME and Windows behave.
BUG: 416025
BUG: 470538
BUG: 421869
BUG: 312027
|
|
This commit extends the dolphinmainwindowtest for the accessible tree
to test that each object a user can reach through the Tab key or the
Shift+Tab key combination is distinguishable by its accessible info
from the previous object.
Objects are considered distinguishable if they have different
accessible names or their first ancestor with an accessible name is
different.
|
|
Pressing Escape on the location bar while in breadcrumb mode has no
effect at all. This commit changes this to instead move the focus
to the active view. This is more logical because a user pressing
escape while having focus on the location bar is no longer
interested in interacting and changing the location in the location
bar. They most likely want to act on the current location instead.
|
|
|
|
Existing settings are converted.
Works on most FS except FAT/exFAT which fallback to .directory files.
If the extended file attributes (in ADS in Windows) can't be saved, they are saved to file as before.
BUG: 322922
You can see file xattr using for instance for Unix filesystems:
getfattr -d /home/meven
|
|
The dolphinui.rc and dolphinuiforphones.rc file describe parts of
the desktop and phone user interfaces for Dolphin. They are meant
to be different in specific ways, but identical in many others.
This test makes sure that the two ui files stay mostly identical by
requiring any differences between them need to be explicitly
stated/registered as an exception in the test. This way no
accidental divergence of user interfaces can happen.
|
|
This commit implements an action to move focus to the Places panel
analogous to "Focus Terminal Panel" functionality-wise.
The implementation of the "Focus Terminal Panel" and "Focus Places
Panel" actions is streamlined while improving their code quality.
The "Focus Terminal Panel" action is moved into the "Show Panels"
sub-menu because it makes more sense to be there considering that its
previous location (the "Tools" menu) is meant for external applications
and not for functionality internal to Dolphin.
This commit also makes it so the keyboard focus is moved to and from
the Places panel whenever it is toggled visible or invisible. This is
now consistent with the focus handling when the Terminal panel is shown
or hidden.
The "Focus Places Panel" is one of the actions which was wished for in
KDE's accessibility chat room because people relying on keyboard
controls might need to press the Tab key a lot to move from the view to
the Places panel.
The new default shortcut is Ctrl+P.
|
|
Add test for double-click activation.
BUG: 485295
|
|
Items should only be selected if the user wants to act on them.
However, previous to this commit we sometimes selected items even
when there is no reason to assume that the user would like to act
on them. Such selections are dangerous because they make it more
likely that the user manipulates items by accident which they
never even explicitly selected.
Example: The "Up" action is used to navigate to the parent folder.
This will implicitly select the folder one emerged from after
opening the parent folder, so just one accidental press of the
Delete key will lead to data loss if the press goes unnoticed. This
scenario would have been avoided if no folder had been selected
automatically.
The above example becomes even more dangerous if the user is acting
with elevated privileges.
The following implicit selections of items are being removed:
- Selecting items that are being activated
- Selecting folders one emerges from
Even though these items will no longer be selected after these
actions, they will still be marked as current.
The only downside I see is that our indication of which item is "current" is a lot weaker than the selection highlight, so it might be more difficult to spot which folder one has emerged from. However, this could be counter-acted with some other temporary indication if this really turns out to be a problem.
The only downside I see is that our indication of which item is
"current" is a lot weaker than the selection highlight, so it might be
more difficult to spot which folder one has emerged from. However, this
could be counter-acted with some other temporary indication if this
really turns out to be a problem.
BUG: 424723
|
|
This MR fixes some issues related to RTL scripts:
- wrong layout in Compact View mode
- broken horizontal scrolling in Icon View and Details View modes
- broken navigation with left and right arrow keys in Details View mode
BUG: 484012
BUG: 449493
|
|
Prior to this commit the test failed because it expected a generic "inode-directory" icon for directories like "home" or "tmp" even though we have more specialised and nicer icons for these directories. I assume the test only used to pass because we were actually always using generic and therefore unhelpful icons for tabs.
This commit removes the hard-coded expectation of the "inode-directory" icon and instead compares the tab icon with the return value of KIO::iconNameForUrl(tabUrl).
|
|
When a bar is toggled visible this usually happens because the
user might want to use its functionality now. However, if bars
appear without animation or at a location the user is not
currently looking at, they might not notice that they have appeared
at all.
An animation makes it more likely that the user notices the change
and can then use the newly made visible component.
Another reason for animations for showing or hiding of components
is that it can be disorienting for users when panels or bars
suddenly appear or disappear without animation. There is no visible
movement then, so the user might not know what happened if they
didn't concentrate or blink at that moment. The newly appearing or
disappearing component might also displace other components which
can make it difficult to find what one was just looking at.
These bars animate now after this change:
- Search panel
- Filter bar
- Status bar
This is implemented by extracting the animation code from
SelectionMode::TopBar into a new abstract base class
AnimatedHeightWidget. This class is now also used in
SelectionMode::BottomBar and the animating code there was removed.
These bars are left in Dolphin that stay without animation:
- Menu bar (Would probably need to be implemented in KXmlGui)
- Tool bar (Would probably need to be implemented in KXmlGui)
- Tab bar (Needs a different appraoch because it already inherits
QTabBar and therefore can not inherit AnimatedHeightWidget)
|
|
Prior to this commit pressing Tab repeatedly would bring the focus to
the end of the status bar but not further.
This commit makes sure the tab focus doesn't get stuck on the invisible tab
bar by explicitly removing the DolphinTabBar from the focus chain while it
is hidden. I don't understand why pressing Tab doesn't do anything for
the invisible tab bar, but removing an invisible and currently useless
widget from the focus chain seems sensible in any case.
Improve the accessibility autotest to prevent regressions concerning this.
|
|
Normally, the session is only saved after a few seconds. The autotest
waits for that to happen. This commit reduces the time until the session
is saved to a fraction of a seconds which means that the autotest will
complete faster.
|
|
This commit reuses our test cases for keyboard navigation to
also test the right-to-left layout direction under various conditions.
There is also a small change to make the error output more
human-readable.
|
|
Adds a simple test to check if session is autosaved when a new tab is
opened.
|
|
Proposed by krop on #kde-devel
Unless my bisect skills failed broken by https://gitlab.kitware.com/cmake/cmake/-/commit/ebc9e448b38378cc21a6b77df03a77d9732c24ac
|
|
compat code
Amends 3f92ac800c5e52d0a5622367ff8ffe8cbbdfa184
|
|
After refreshing the view when
size count is selected as the sortRole,
count is 0 at first. When the actual count is loaded,
the sorting is done according to the count being 0.
This can break the sort order and cause view and model
to be out of sync.
Making sure we always resort all items
when the directory size/item count is changed fixes this
BUG:473999
|
|
Remove qt5 specific cmake code too
|
|
|
|
This is to make sure that all items that can be tabbed to and their
accessible children have the main window somewhere in their
ancestry/ancestor tree.
|
|
Before this commit, KItemListViewAccessible would always return
nullptr as its parent. This meant that accessibility software would
have to guess to which window/hierarchy the KItemListView belongs
to. Guessing shouldn't be necessary here.
This commit makes sure that the KItemListView always returns a
sensible parent in the accessible hierarchy. It does so by
explicitly setting the accessible parent for every KItemListView
after construction in the DolphinView contructor. Since
KItemListView now always knows about its accessible parent, the
accessibleInterfaceFactory can always ask the KItemListView for
that information when constructing the QAccessibleInterfaces.
Fixes https://invent.kde.org/system/dolphin/-/issues/47.
|
|
|
|
|
|
|
|
|
|
|
|
CCBUG: 196772
|
|
|
|
|
|
This reverts commit df5cf6fe851021a73159fdd3487fe629c1ea8376.
The autotest was orginally changed to adapt to new behaviour when
the Space key is pressed. This change used to be implemented in
KItemListController which meant that the KItemListControllerTest
needed changing, but this is no longer the case.
Instead the new behaviour when the Space key is pressed is now
implemented further up the hierarchy, so the KItemListController
change could be reverted in
7b6a67e520c04f56b4b05fa26b252177398df6df and as such we can also
revert this autotest change.
|
|
The basic mainWindow methods openFiles(), openDirectories(),
isUrlOpen(), and isItemVisibleInAnyView() have some non-obvious
behaviours that shouldn't regress. We never want to open
directories that are already open. We only want to open new tabs
when the currently open ones don't have the file that should be
highlighted in view already. This should work even if view modes
that allow folder expansion are in play. It should also work when
the view area is split.
|
|
|
|
|
|
The old expected behaviour was that pressing Space would select
the current item. We now trigger selection mode instead and don't
select. At some point in the future we might want to instead have
Space trigger a "QuickLook" feature for quick viewing of full
files. In any case, the old behaviour of having Space select is
no longer expected, but Ctrl+Space can still be used for this.
It might also make sense to have Space trigger the old selection
behaviour if both selection mode and "QuickLook" are disabled, but
that is very low priority and will cause more code complexity than
the non-default benefits are worth.
|
|
If current directory is a local file, try to find nearest dir ancestor and
open it. Display warning to the user.
|
|
Improve the test by also testing the various selection mode bars.
No combination of bars should lead to a resize of the paces panel.
This commit also helps with increasing code coverage by making all
possible selection mode bars appear.
|
|
This merge request fixes up a little oversight by me. I will merge this as soon as the pipeline passes.
------------------
Some lines verified that nothing is selected when navigating to a
folder that was not acted on yet. These verifications didn't test
anything meaningful because the folder in question was empty.
This commit adds a file and a folder to the test folder so that
testing if nothing is selected means something.
|
|
- The first test makes sure that the width of the places panel
doesn't change no matter what other panels are shown or hidden.
There used to be bugs about this.
- The second test is kind of an integration test. It makes sure
that the go actions work correctly. On two occasions we had
the regression here that after going "Up" in the file system
hierarchy the folder one emerged from didn't have keyboard
focus which makes peeking into multiple folders tiresome.
The test also makes sure that going back and forward works
as expected and there are some sanity checks about which
go actions are enabled, which items are selected and that
using tabs doesn't interfere with any of that.
|
|
- Fix kio_version.h include
- Explicitly create a QSet for QCOMPARE
|
|
If we have directory "a" and "a/b" and expand both, then collapse "a" we
tell KDirWatcher to stop watching both these directories.
However, KDirWatcher keeps them in the listersCurrentlyHolding list as
well as the listersCurrentlyListing list and will still notify of
changes. If a new file appears in "a/b/" we will still get change
notifications.
When dolphin processes these changes we cannot find the relevant parent
node. It then gets confused and inserts the item into the root directory
from the POV of the model notifications. When we then open the relevant
folder the model knows a node with that URL exists and fails to add it
correctly.
This can also be reproduced by continually downloading files into a
subdirectory tree and rapidly expanding and collapsing folders a few
levels deep.
|
|
|
|
|
|
They were used by the custom places panel and are now unused.
|
