https://puri.sm/posts/librem5-progress-report-6/

Android running apps design is broken pace of garbage.

But purism 3 seconds to access running apps must be the most epic fail ever.

They have to fix it! I want same way as SFOS and Harmattan.

On this one, I agree with you.

The whole gesture interface they have proposed seems broken, not only the 3s multi-tasking view.

Too bad they don't seem to have read what Jaakko Roppola explained a long time ago (and that even Jolla broke with Sailfish 2, in place of fixing what was missing in Sailfish 1...), like this article:
https://jaakkoroppola.blogspot.fr/20...-gestures.html

First, the interface is not symmetric, so right handed and left handed person would not have the same actions easily accessible.

Then, as I see it, the most important function in gestures would be the return to home screen or multitasking view depending on the OS design principles (would be multitasking on N9, home screen on Android or Symbian for widget access, kind of a combo on Sailfish).
As the most fundamental action, this should have been an edge swipe. But no, the upper right edge swipe is "most frequently settings". And they have hidden the multitasking action behind a 3 seconds delay on the bottom and the homescreen action on the lower swipe.

The secondary most important action would be in my mind the events view. Any well designed OS uses this event view to provide all kind of informations about communications, calendar, etc. and is used to see the details and eventually directly reaching the dedicated application to go further. That is a lot more powerful then launching individual apps and navigating through them to find the info. Following Jaakko principles, this would be on the lower swipe. Here, they put it on the worst place possible for a right handed person: top left edge swipe...

For the rest of the actions, I can say the quick launcher idea can be interesting, but the home-screen role should not be that different to need a different concept.

The multi-tasking view as a carousel they show has the advantage to be easily clickable from about any point on the screen, whereas tiles (like in the sailfish 3x3 grid) can be hard to reach if too small and to far from the thumb rest position (like the top left on sailfish for right handed persons when there are 3 columns). The drawback is that it is harder to quickly spot a specific application if you have a bunch running (as would be expected from a real multitasking phone).

And last, I don't know in all of that how you can start a new application. There is no app launcher ? Would this be on the homescreen ? Jolla made the mistake to put it on the bottom swipe in place of the event view (that is now hidden behind the homescreen) as the second most important action, where it is secondary you mainly are using apps already launched or opening them in reaction to notifications. But it should at least be somewhere...

So far I liked what they where proposing, but this UI post left me wondering about what this will look like in terms of usability out of the box.

Looks just awful. Looks like somebody had just printed a picture of a screen and then started to draw this picture like "let's put this action here" without even thinking how usable it'd be. In Roppola's slides there is good guidelines on what actions are most important in mobile OS and what kind of swipes are easy to do, with both hands and so. Jolla's final concept isn't exactly as innovative, but still usable on both hands. Those swipes presented in that Purism post are just unreachable and stupid. Like a desperate need of being different, just because of being different. I hope they realize it themselves before it is too late.

This is the same silly
" re-invent everything while designing new hardware" blunder
that torpedoed Ubuntu (and others?).

The Ubuntu team decided to reinvent the browser, email client,
media player and fundamentally rewrite everything
instead of simply port linux onto a handset.
That debacle sank in an excruciatingly slow demise.

Let us sincerely hope they just make a handset we can boot USB
and install whatever Debian/Gentoo/Arch system we want.

Beyond question whatever new and exciting UIX they design
will be greatly received by some and quietly disputed by many.

Better to also support the basic linux ecosystem
which already has globally 'baked in' support.

A more current example if is the very excellent GPD Pocket.
Crude, but it is an excellent success of hardware
hacked together which does support whatever Linux
you care to run on it.
The OS it ships with is 2nd best but who cares - the hardware works !

Actually what I'd like to have is a swipe-gesture interface that would be completely user-definable.

A mobile device is after all a personal device, no two people should be forced to use same gestures to operate their device. On startup there should be some sensible defaults set, but then everything should be personificable. There is some minimal tuning available in SFOS, for example turning off the foreground application closing by downswipe but that is not nearly enough.

I guess a problem with that would be the extra burden on the support department: people complaining "it doesn't work", because they (or someone else) changed something and can't remember it, or can't figure it out anymore. Technical aptitude isn't as common as you might hope in this day and age.

Don't forget they are not designing a new N900 with a keyboard and resistive touch panel that can be used to precisely hit desktop like buttons using a stylus.
The shell must be at least usable with a capacitive touch only, which is where gestures are the best, and the basic applications for a phone should be available as touch-friendly.

I think we agree on this one. I have nothing against being able to configure it, and my last sentence was:
And out of the box is precisely what this blog post was about, in their own words:

While on their blog to take the final quote, I found the found UI progress report that I missed when it was published: https://puri.sm/posts/librem5-progress-report-4/

There are more details on the homescreen at the end of the post. So I have now found where the application launcher is ! It is below the homescreen, a bit like on Sailfish OS 1.0.

And interestingly, there is this sentence: "Less recurrent gestures, like accessing the settings or the detailed list of notifications, are available in the upper part of the screen."

So I guess I found where a part of our disagreement in the UI comes from !

Time for weekly update a little late this time but...

https://puri.sm/posts/librem5-progress-report-7/

I really don't get why they try reinvent the wheel when there already exists good UI libraries and wayland shells/input methods!?!?!?

Seriously they have no chance release something stable 2019 if they going reinvent everything!

They should use Qt since it already is has most touch friendly UI out there. Latest 3/4 QT versions has swipe/touch friendly components so why on earth redo it!?

https://doc.qt.io/qt-5/qtquickcontro....html#examples
https://doc.qt.io/qt-5/qml-qtquick-c...swipeview.html

They should concentrate on the backend stuff and throw shell on the trash and do it in Qt using qt quick controls. We don't ned just another wierdo SDK.

Continue using Qt would make it alot easier for SFOS/Harmattan/Nemo/Android developers too.

Yes I know this is prof of concept but still. Why use time in some prof of concept when they instead should concentrate make latest Qt compiled and installed on iMX? Now that would much cooler to see some real UI than some wierdo hack..

Exactly, three questions I'd like Purism to answer
1. What's wrong with Qt Quick? QtQuick Controls 2 are meant especially for mobile and are developed for a really long time?
2. What's wrong with whatever compositor Plasma Mobile/SailfishOS/Nemo Mobile use?
3. What's wrong with apt/yum/zypper/pacman or whatever Alpine uses that they want to go for flatpak? Why O(n) copies of libstdc++? (if my understanding of flatpak is correct) Who is going to update the dependencies when there's a CVE?