| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This work removes the manually curated lis of contributors, both because
of its maintenership burden and because contributions come from a very
diverse sources, including marketing, QA, translation, design,
mouth-to-mouth preaching, etc. By creditting just a few of them we
undermine all the others that make such a great and sometimes invisible
effort.
This also updates the program name to mention the codename Nautilus, so
those looking at the about dialog, most likely developers, know the code
name of the project.
And finally, it adds the website so people know where to find more
information, issues, and to contribute.
|
| |
|
|
|
|
|
|
|
| |
destroy_conflicts_task_data() frees the task data struct before clearing
the mutex and conditional variable inside it.
Addresses https://gitlab.gnome.org/GNOME/nautilus/issues/628
|
|
|
|
|
|
| |
run_script() has a debug print, which references a freed string.
Addresses https://gitlab.gnome.org/GNOME/nautilus/issues/628
|
|
|
|
|
|
|
|
|
|
|
|
| |
After a40f2b5415b2d919be28bf19e29c4ea840f87685, the button data is only
removed when button_data_free() gets called explicitly, which only
happens in two places: nautilus_path_bar_clear_buttons() and
button_data_file_changed(). What also happens is that buttons are
removed as part of their containing container destruction, which was
accounted for when a weak ref was used. Now that that never happens,
some leftover signal handlers wreak havoc.
Fixes https://gitlab.gnome.org/GNOME/nautilus/issues/593
|
|
|
|
| |
Unused since commit 4c6d4eea995a66e80fb21ee83b66954d1ce71c08
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since commit 4c6d4eea995a66e80fb21ee83b66954d1ce71c08, we haven't been
increasing the reference count of the recent search engine when not
creating a thread.
But the idle callback unref's it regardless, which causes the recent
engine object to be destroyed while NautilusSearchEngine still holds
a reference to it. This causes a crash when trying to stop the
already destroyed recent engine.
Restore reference count increase to fix this.
Fixes https://gitlab.gnome.org/GNOME/nautilus/issues/692
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 0009d0e771f1a7cb9adf12e78d66aefcccc766b6
That commit gave the pathbar a bounding frame under HighContrast.
However, this 1px border is not consistent with button borders, and
it lacks padding. The result is uglier than before the change.
Furthermore, it may be the case that setting GTK_STYLE_CLASS_BACKGROUND
and GTK_STYLE_CLASS_FRAME in a GtkBox is either meaningless or leads to
unexpected styling.
This change is neutral for Adwaita, so let's revert it.
Fixes https://gitlab.gnome.org/GNOME/nautilus/issues/674
|
|
|
|
|
|
|
|
|
|
| |
We used to emit the "focus-view" in handle_event() when the [Down] and
[Up] keys were pressed.
But this doesn't work after the previous commit.
To restore it's function, use the "focus-view" signal into a key
binding signal.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit dae54e0c72e92a3129c4e8754bd9f3025e487c72 made the slot handle
the key press event before chaining up.
But this breaks the [~] and [/] accelerators, possibly because the slot
handles them as type-to-search, stopping the event without chaining up.
To fix this regression, revert to chaining up first. Pressing [Down] on
search entry to focus the view won't work like this, but we are fixing
that in the next commit.
Fixes: https://gitlab.gnome.org/GNOME/nautilus/issues/684
|
| |
|
|
|
|
| |
Wrong path for the html file...
|
| |
|
|
|
|
|
| |
And show it in GitLab Pages. This will help us to keep an eye on the
status of testing. The report can still be improved though.
|
|
|
|
| |
So they are more equal and we have a simple way to generate it.
|
|
|
|
|
|
|
| |
So we can generate coverage reports.
And do it with pip, cannot make my system run the pipToFlatpak tool
locally.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
This is necessary for files such as GDrive files, where the target uri
is the web url that needs to be opened for activation.
Closes: https://gitlab.gnome.org/GNOME/nautilus/issues/473
|
|
|
|
|
| |
Since we only care about whether those macros are defined, we can use
set to save a few lines.
|
|
|
|
|
|
| |
A number of CPU architectures are unsupported as far as seccomp goes.
This commit expands the current restriction to Linux by checking the
host CPU.
|
|
|
|
|
|
|
|
|
|
| |
Setting LANG and NO_AT_BRIDGE will prevent warnings about LANG not being
set (obviously) (Tracker), and about the a11y bus not being up, which we
don’t necessarily need to be able to test things.
Starting a D-Bus session before running tests ensures that dbus-launch
doesn’t croak when Tracker tries to reach the SPARQL backend, because of
missing X11 support.
|
|
|
|
|
| |
If, for some reason, the cursor is null, we end up with a critical when
unreffing. Not very good when running tests.
|
|
|
|
|
|
| |
tracker_notifier_new() may return a null pointer on error, and that’s
exactly what happens sometimes (CI pipelines,
https://gitlab.gnome.org/GNOME/tracker/issues/43).
|
| |
|
| |
|
|
|
|
| |
It’s no longer an enum, but rather a string.
|
| |
|
|
|
|
|
| |
a9c446afe262b7989bc17ef1f9ab19931db8dede inadvertently removed it and
that is causing leaks that don’t make sense at first.
|
|
|
|
|
| |
filtering_changed_callback() populates a list of directories, but never
frees the list.
|
|
|
| |
The ID was changed to reflect that in the manifest that is built.
|
|
|
|
|
| |
Use a new function to initialize it without having to redo the same
thing twice.
|
|
|
|
| |
RecentInfo short-name needs to be free'd
|
|
|
|
|
|
|
| |
Since this is already part of the search engine struct, and given
that we ref it, there's no point of reffing the query again.
Also not to create a mem-leak when we don't run it.
|
| |
|
|
|
|
|
|
|
|
|
| |
Since these are forwarded to the title bar, in cases where the middle
click does something special (e.g. minimize the window), unintended
side-effects might be observed in conjunction to opening the location in
a new tab. Claiming the sequence helps prevent that.
Fixes https://gitlab.gnome.org/GNOME/nautilus/issues/671
|
|
|
| |
Closes https://gitlab.gnome.org/GNOME/nautilus/issues/395
|
|
|
|
|
| |
Some (public) functions needlessly assert certain facts about their
parameters, and can just throw warnings and return early, instead.
|
|
|
|
|
|
|
|
|
|
| |
Cloning the window while its location is null (i.e. the attributes of
the pending location file are still being waited on) results in a failed
assertion, which is less than desirable. That can be fixed by
preemptively checking if there is a pending location on the slot and
using that in the new window.
Fixes https://gitlab.gnome.org/GNOME/nautilus/issues/649
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On wider windows, we show a border around the pathbar and fill it
with background color using custom CSS specific for Adwaita.
This means the pathbar is unbounded in HighContrast mode or 3rd party
custom themes, giving the headerbar an uncentered, unbalanced layout.
Fix that using the "frame" and "background" style class which are
exposed by Gtk+ for this purpose.
Closes: https://gitlab.gnome.org/GNOME/nautilus/issues/661
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a regression from 3.28, where the file items where being
selected while searching if the user clicked either up or the down
arrows.
Since 3.30 we moved the query editor to the header bar, this automatic
handling was lost.
To fix it, do something similar as we do when activating the search,
which is emitting a signal to inform the window slot and the views to
perform an explicit action. In this case, we focus the views so further
key events are handled by the views themselves and not by the query.
Closes: https://gitlab.gnome.org/GNOME/nautilus/issues/610
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We use custom CSS that undecorates pathbar buttons in Adwaita. However,
this doesn't apply to HighContrast, which keeps button borders, and it
looks bad because there are slashes between them.
While some custom styling is still needed, some can be replaced using
the public "flat" class, with the benefit of removing the button
borders in HighContrast (as well as any 3rd party custom theme that
supports the "flat" class).
Do that. Also, remove the now pointless "linked" class.
Fixes: https://gitlab.gnome.org/GNOME/nautilus/issues/661
|
|
|
|
| |
Unused since commit b1abb52610dbbcac3669347663459e4b00241d57.
|
|
|
|
|
| |
I'm not gonna say how much of the day I spent on this code because
I read it the other way around because of the misleading function name.
|
|
|
|
|
|
|
|
|
|
|
| |
Querying the file cache in is_recursive_search() does not guarantee that
the file system field will be populated with a value, and somehow
results in data races, in turn resulting in random crashes/file list
corruption.
Hopefully fixes https://gitlab.gnome.org/GNOME/nautilus/issues/640,
closes https://gitlab.gnome.org/GNOME/nautilus/issues/632 and resolves
https://gitlab.gnome.org/GNOME/nautilus/issues/602
|