automatically adding newly installed applications makes the launcher unusable

[Martin Pitt]

This got spawned from bug 955147. https://wiki.ubuntu.com/SoftwareCenter/LaunchingApplications decided to add newly installed applications to the launcher by default.

This new concept essentially admits defeat that new users do not know about the dash. This started last cycle with moving the control-center icon into the launcher (thereby making it overflow on netbook screens), which already spawned a large discussion. Now this proposes to throw all limits over board and add tons of new icons to the launcher; NB that we make it very easy to install lots of new apps, and we go through great lengths to make it possible to create, find, and install third-party apps even post-release.

Now, this approach works around a temporary 5 minute question ("where does that app go"?), not by giving a 5-minute answer (quick tutorial or hint application which opens up the first couple of times), but by essentially redefining what the launcher is: a place where all applications go. This has several problems:

 * Our launcher just isn't built for that -- it's efficient and useful for up to 10 to 20 icons, depending on your screen size. But after that it folds, and exponentially gets harder to use. It doesn't have text to describe the applications, does not have a stable order, and is just one-dimensional.

 * It does not have most of the applications that are installed by default -- how do people find that?

 * It teachs the wrong thing -- that the launcher is the place for all apps. It drives people further away from the dash, thereby aggravating the learning problem instead of solving it.

 * It optimizes a thing which represents 0.00005% of the workload [1] at the expense of making the other 99.99995% much worse.

The spec itself points out that this "relies on heavy users removing unwanted launchers". Given how the launcher works, anyone who installs more than ~ 10 apps is a "heavy user", i. e. quite a large part of users.

The flaw in that spec is the conclusion it draws from the fact that users are not able to figure out how the dash works: You rely on users being able to remove apps again, but then they would again need to use the dash to start these applications, a capability you just denied them.

So in summary, this turns the launcher into something like the app launcher that e. g. Android has -- place for all application. But as it's not built for this and there are no improvements for at least mitigating the effects of this change.

Making the launcher useful to hold an arbitrary number of apps requires it to become two-dimensional, sorted, searchable, and touch friendly. We already have that -- the dash. So if we want to go that direction, could we just not use the dash as we have it, and instead change how to invoke it?

[1] Assuming a very pessimistic scenario: using Ubuntu for two hours a day for a year, and 50% not finding out how the dash works in 5 minutes.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Omer Akram (om26er)]

This bug is in need of feedback from design so as per Unity bugs convention the status goes to incomplete. Thanks for the bug report.

[Matthew Paul Thomas (mpt)]

There are several ways the Dash could be improved to increase visibility of new applications:
(a) Moving the Dash button out of the launcher would make it visible even when the launcher is hidden. (Previously in the menu bar it was not noticeable enough, but I think that was merely a visual design problem.)
(b) The Dash home screen could show recently installed applications under "Recent Apps".
(c) The button for the applications screen could glow when there are newly-installed applications.
(d) Newly-installed applications could similarly glow inside the applications screen.
(e) The applications screen could stop displaying applications that aren't installed, making clearer which ones are installed (and solving other problems too).

And there are several ways the launcher could be improved to more gracefully handle added items:
(f) Placing it horizontally instead of vertically (or even just providing this option) would give it more room.
(g) Moving the Dash button out of the launcher would leave more room for other things.
(h) Making the workspace switcher tile optional, and off by default, would leave more room for other things.
(i) Switching from accordion folding to something that retains equal access to all items, or something that provides a simple overflow menu, would more gracefully degrade with many applications.
(j) Dragging an application directly to the Trash could remove it from the launcher, instead of uselessly temporarily scrolling the launcher.
(k) "Unlock From Launcher" could be reworded to something more obvious, like "Remove From Launcher" or "Remove From Launcher When Closed" (depending on whether the application is currently open).

But even if the Dash was improved in all those ways, and the launcher was improved in none of those ways, I would still think that newly-installed applications should go into the launcher by default. The launcher is visible most of the time, the Dash is not, and that's a huge handicap for the Dash to solve the problem of finding newly-installed applications.

[David Barth (dbarth)]

The problem can't be solved for 12.04. It's definitely 12.10 material.

[Martin Pitt (pitti)]

> The launcher is visible most of the time, the Dash is not, and
> that's a huge handicap for the Dash to solve the problem of finding
> newly-installed applications.

I find it a little disconcerting that this point is brought up over
and over again without any response to any of the problems that this
creates.

[Martin Pitt (pitti)]

> The problem can't be solved for 12.04. It's definitely 12.10 material.

Reopening then. This is not a precise task, it's the "floating" task.

[Martin Pitt (pitti)]

John, can you please explain why this is "Invalid"? I'd accept "Opinion" (but would then appreciate some explanation and responses to the points raised in the description), but closing this as invalid without any explanation feels pretty harsh to me.

Thanks!

[John Lea (johnlea)]

Hi Pitti, I marked this bug as "Won't Fix" in Unity. In the "unity (Ubuntu)" distribution the "Won't Fix" option was not available to me, so because the bug was "Won't Fix" in Unity the closest status I could give it in "unity (Ubuntu)" was "invalid". Re. the status itself, I am currently cleaning up all the design bugs before the PS sprint, and rather than leaving this bug in limbo it is better to give it a honest status which at the moment which is "Won't Fix". Sorry!