An update on current stability problems - was: Re: Problem installing in Windows

[Stefan Rödiger]

Hi Thomas,

Great work. I have just downloaded and (slightly) tested it. No issues, no 
crash.

Kind regrads
S

On Samstag, 14. März 2026 13:32:59 Mitteleuropäische Normalzeit Thomas 
Friedrichsmeier wrote:
> Another update:
> 
> I'm relieved to announce that our development binaries should now be
> functional again, at least the AppImage and the Windows builds. The "fix"
> is a combination of updating to Qt 6.10.2 and a new option to disable
> QWebEngine hardware acceleration (on by default).
> 
> I've also made some progress on providing optional rendering using QWebView,
> but that still needs quite some work (not part of any current builds).
> 
> Regards
> Thomas
> 
> Am Sat, 7 Feb 2026 11:34:35 +0100
> 
> schrieb Thomas Friedrichsmeier <thomas.friedrichsmeier at kdemail.net>:
> > Hi,
> > 
> > a short update on my thoughts/plans to address current -
> > QWebEngine-related - stability problems, as this may actually take some
> > time to implement:
> > 
> > 1. It feels like 90% of our issues stem from QWebEngine, of late. And this
> > is not even counting problems WRT building and packaging.
> > 
> > 2. QWebEngine 6.10.1 - which is currently in use on our builders - appears
> > to be a particularly buggy version. Qt 6.10.2 is out, and there is some
> > hope that it will resolve some of the issues, but I cannot give you a
> > timeline on just when it will be available for our builders (hopefully
> > within a week or two), nor a promise that it will actually help.
> > 
> > 3. Even if QWebEngine 6.10.2 does solve the current problems, I'd love to
> > have some alternatives. Yesterday I learned that Qt's other web component
> > (QWebView) has meanwhile learned to use system native HTML viewers on
> > both Mac and Windows. That sounds like it may avoid a whole class of
> > troubles at least on those platforms. QWebView is not going to be quite
> > trivial to use, however, as the coupling is much more indirect, and we
> > will have to somehow add support certain functionality (importantly
> > handling rkward:// and help:// urls) via injected javascript, and -
> > probably - websockets. I'll experiment in that area, next, hoping this
> > will turn out to be a viable solution.
> > 
> > 4. In the longer term, I hope Servo will bring back some diversity into
> > the
> > embeddable HTML render landscape. But that does not look like a solution
> > for 2026, yet.
> > 
> > Regards
> > Thomas