Due to somewhat confusing variable naming, we were accidentally setting
m_initializingBackends to false, when that variable actually holds the
number of backends that are still initializing. Instead we need to set
m_isInitializing to false.

This fixes a bug where the placeholder message for the MIME type screen
would fail to appear as expected, because one of the conditions in its
opacity property was `!ResourcesModel.isInitializing` and that was never
getting set to false.

BUG: 457528
FIXED-IN: 5.24.7

Check whether the reason that the !ResourcesProxyModel.isBusy is that
the backends aren't all populated yet.

This way we make sure we don't get a flash of the error message in the
placeholder.

Minor build system fixes, container sizes, QVariant comparison, missing
includes, QStringRef.

Aleix Pol Gonzalez authored Feb 01, 2022 and Aleix Pol Gonzalez committed Feb 21, 2022

BUG: 449259

(cherry picked from commit 9476b7b2)

Makes it easier for us to interact with it from QML as QIcon is not even
a Q_GADGET and we can't query its information.
Also use the iconName when using the DiscoverAction as a
Kirigami.Action.

At least in English, the word is shown no fewer than 8 times. This
commit brings that number down to 6, which should be a bit less
ridiculous.

Nate Graham authored Sep 22, 2021 and Aleix Pol Gonzalez committed Sep 23, 2021

Discover currently tells users to file bug reports against their distro
in a few places, but doesn't tell users how to do so, and in any event
is often is bit cagey about what exactly the error is.

As a result, we still often get bug reports against Discover that are
distro issues which we can't do anything about. This wastes developer
and bug triager time in KDE, frustrates users by giving them the
run-around, and makes Discover look bad in cases where the problem is
someone else's fault.

This commit attempts to improve that situation by making the error
messages clearer and also adding a big "Report this issue" button that
users can click on which will take them to the bug report URL for their
distro, not for Discover itself.

BUG: 442785
FIXED-IN: 5.24

Nate Graham authored Apr 07, 2021 and Aleix Pol Gonzalez committed Apr 08, 2021

This tooltip was added to help diagnose issues with stuck backends, but
has proven troublesome. This commit removes it for the following
reasons:

1. It did not succeed in fixing the problem in the first place because
   the root cause always seems to be the PackageKit backend being stuck
   for a moment and not issuing updates, so the tooltip doesn't change at
   all or even tell us which backend is stuck.
2. Due to a bug, the tooltip was always shown when the cursor was
   hovered anywhere over the main panel, so users saw it much more than
   we wanted and filed many bug reports.

This commit partially reverts 2196ed49.

BUG: 432044
FIXED-IN: 5.21.5

If you want git blame to ignore formatting revisions run:
git config blame.ignoreRevsFile .git-blame-ignore-revs

Aleix Pol Gonzalez authored Feb 05, 2021 and Aleix Pol Gonzalez committed Feb 08, 2021

It's big part of the reason why we use QtWidgets and it seems to add
some unnecessary overhead.

Fix annoying warning when connecting to a nullptr emitter

Adds a tooltip to the "Fetching updates..." label so that users can know
what's holding the update back.

Keeping it in a tooltip since it's something that never should be seen,
except when there's a problem somewhere in the system.

It's not used elsewhere and skips the creation of a weird intermediary
container.

Try not to keep objects that might get destroyed in memory unless necessary

CCBUG: 412632

Summary:
Creates components so it can be integrated in any Kirigami application though.
Haven't tested custom information providers yet, need to get hold of a server
first to see how it will play out.

Depends on D24010
Depends on D24011

Reviewers: #plasma, vkrause, ngraham

Reviewed By: ngraham

Subscribers: ngraham, mart, plasma-devel

Tags: #plasma

Differential Revision: https://phabricator.kde.org/D5961

Summary: It takes longtime on some distros, so we better show some information about how it's evolving.

Test Plan: Tested on KDE Neon, which is where I've suffered this most

Reviewers: ngraham

Reviewed By: ngraham

Subscribers: plasma-devel

Tags: #plasma

Differential Revision: https://phabricator.kde.org/D23324

To ensure we only have one allInitialized emission (at least just
one per run of the event handler), swap out the singleshot inline
timer for a proper QTimer instance. Also, use this timer elsewhere
the emission was previously done, just in case, and for the sake
of consistency.

Differential Revision: https://phabricator.kde.org/D18262

In case this is in fact the first backend to be added, and also
happens to be pre-filled, we still need for the rest of the
backends to be added before trying to send out the initialized
signal. To ensure this happens, this patch schedules it for the
start of the next run of the event loop, and ensures that it
only happens if there are no more backends to initialise.

Specifically, this happens if e.g. the fwupd backend is the
first to be initialised, and results in (amongst other things)
that the application loads up and insists that there are no
application backends.

Differential Revision: https://phabricator.kde.org/D18246
BUG: 401334

We already have it in ::search() with async API which always gives a a
slightly better API.

This way we won't be getting an overwhelming list of resources at once.

could be the one responsible for the isFetching state

BUG: 389072

Implement pull to refresh on the updates view (although it will only work
on touch screens.
Make the update action the only one that the ResourcesModel offers and make
it globally available

It got watered down over time, doesn't make sense anymore

BUG: 381170