Qt 6.0 Beta 4 Released | Hacker News Reader

Qt 6.0 Beta 4 Released

phoronix.com

82 pointsvarbhat23 comments

I'm eyeing Qt (QML/Quick in particular) as a UI toolkit for a smart TV form factor-like embedded app. Primarily keyboard controlled. Small scale project, just myself initially.

Qt doesn't really seem to focus much on this form factor, but they do have a big win; LG's webOS.

I've found some basic blog posts on the topic but nothing that has really convinced me. Anyone with first-hand experience on how this would work? A lot of upstream swimming?

I used QML for a similar application because it would allow me to assemble the OpenGL scene nicely. I used libMPV to render video frames and QML could easily overlay controls and text. You could even render Chromium (QtWebEngine) on the same scene with minimal effort. I know enough to be dangerous, but not enough to dive deep into the stack of technologies, nonetheless, I got a decent app working with no bugs that customers ever ran into.

For my skill level and effort put forth, I don't know what I could have used to would have worked as well.

Thanks, sounds promising.

QtAV next time :)

I haven't dug deep into QML's keyboard focus stuff but it's been a breeze for developing my photo editing desktop app. It makes gui hacking so fun.

https://filmulator.org

I've worked on a couple of d-pad only uis using QtQuick, Including for a set top box.

Gstreamer's qmlglsink works quite well with zero copy video playback. So CPU usage sits at 5-10% for a nice UI overlaid on top of a 1080p video stream on an embedded system. There are other third-party qt wrappers for gstreamer, but we didn't feel the need for it. Some were outdated too.

https://github.com/GStreamer/gst-plugins-good/tree/master/te...

Those examples show how to do basic video playback using gstreamer.

The key navigation really isn't that difficult either: https://doc.qt.io/qt-5/qtquick-input-focus.html

Just build your UI with a PageStack of Pages consisting of ListViews and GridViews, and that takes care of 90% of your navigation issues without you having to do anything.

For eg.

* filesystem explorer: ListView

* playback controls: GridView

Video Player Page: filesystem explorer + playback controls + video item and the page determines which item should get keyboard focus, based on certain actions/conditions.

So far, the main pain point was licensing. Some of the very useful qt quick components (like layouts for one) are licensed under LGPL v3, so you have to be careful if you're including then in your system - because then you'd have to provide a way for the user to swap out qt libraries or have to go for the commercial version.

Another pain point was CI. We had to manually build our own container with the Qt version we needed. The situation for CI improved recently i think.

This gives me some starting points.

Licensing: Thanks for that warning. I get that they want to make money, primarily off embedded customers. Haven't figured out my licensing model for this project yet, might even go full GPL.

To expand on what I'm misssing: Official/built-in support for spatial navigation, like:

https://github.com/qt-labs/cursormanagement

I'm guessing most Quick widgets aren't immediately usable in this form factor. I'd guess they'd need heavy skinning efforts and custom cursor management to have a chance, and still only the most basic ones would be usable.

Is there any other ~open UI toolkit that's good at TV-style/D-pad-focused navigation out there?

Don't you want a custom design (re "skinning") anyways for an application like that?

Yes, navigation is something you'd need to at least partially build yourself, but in my experience it's not a big deal (I haven't done TV apps specifically, but lots of embedded things with D-pad-like controls), since most UIs have clear navigation hierarchy. QML is very common for this kind of application, and I'm not aware of anything else widely used and open that'd be better for this specific thing.

> Don't you want a custom design (re "skinning") anyways for an application like that?

Yes, probably. I was still holding out hope for some framework that had some decent basic widget designs for a TV-like format so that I didn't have to do all of the work on my own, but I guess that doesn't exist. (I can adjust the design of my app to whatever is available.)

Thanks!

How is the licensing for QT for personal projects?

What does that mean to KDE? Another five years of transition before we get a new stable KDE?

Qt 6 has no major API changes from Qt 5.

I'm no Qt expert but on the linked page it says:

> Qt 6.0 does drop Qt Script and also removes Qt Multimedia / Qt Bluetooth / Qt Virtual-Keyboard modules until a later Qt6 version.

Qt Script was basically never adopted by anyone and was effectively a dead module by the time Qt 5 released, even if it was only officially deprecated in 2015 with Qt 5.5.

Qt Multimedia is actually pretty important, but I assume they have a good plan in place for that.

Not sure about that, plenty of deprecated things removed, plenty of modules removed, plenty of things moved about, ANGLE no longer supported on Windows...

I guess one can debate what constitutes "major" (a bunch of small stuff also adds up), but it certainly looks like it will be a lot less painful than 4->5 was.

So why did they even change the major version in the first place?

> So why did they even change the major version in the first place?

Note that minor versions of Qt are (modulo the occasional bug) API and forward-ABI compatible - an app built against Qt 5.x should still work without being rebuilt when updating dlls to Qt 5.15

This obviously won't work for Qt 6 - there are some API changes, deprecated / removed stuff, in Qt 6 we have `template<typename T> using QVector = QList<T>;`...

One more major change is that raises c++ compiler requirements to c++17.

It says that they did drop some APIs.

> Qt 6.0 does drop Qt Script and also removes Qt Multimedia / Qt Bluetooth / Qt Virtual-Keyboard modules until a later Qt6 version.

So I guess it is major if you use those APIs.

Not everybody follows semantic versioning.

Qt does, though - it breaks ABI and API compatibility with major releases, but the last big change was with Qt4. 5 and 6 are / will be easy to port to from earlier versions.

Consider that removing any public function / method / class is an API and ABI break. Qt does do these with major releases, but how much can vary considerably from version transition to version transition.

Reply on news.ycombinator.com