Review Request 116648: Split into one KCM for Desktop Effects and one for Compositing

Martin Gräßlin mgraesslin at kde.org
Mon Mar 10 10:12:12 UTC 2014



> On March 7, 2014, 4:15 p.m., Hugo Pereira Da Costa wrote:
> > Hi Martin,
> > Any idea why the background behind the shells is flat ? Does it happend with all KCMs ?
> > (I'll double check asap but right now my systemsettings is crashing because my kauth is fubared. Links against incorrect pluggin, and qt4 and blows)
> 
> Thomas Lübking wrote:
>     Known QML issue (at least I bet this is)
> 
> Martin Gräßlin wrote:
>     clearly QtQuick issue
> 
> David Edmundson wrote:
>     Yes and no.
>     We're quite clearly hardcoding a solid background colour here:
>     Rectangle {
>         color: engine.backgroundViewColor()
>     }
>     
>     If QtQuickControls was working it wouldn't fix that.
>     But QtQuickControls doesn't handle the radial background yet, so it's kinda a moot point.
>     
>     It /might/ be possible to not set a background here, and set the QQuickView QWindow to be transparent and we'd see the radial background of the underlying KCM QWidget?
>
> 
> Thomas Lübking wrote:
>     According to the findings in some other RR: no.
>     QtQuick/QML is designed to be some scene (graph driven) and esp. if it uses HW acceleration, merging in the widget backingstore should pretty much defeat that approach - a solution in that direction (even on top of a Q*View patch) should be pretty "ugly" (rendering all widgets below into an image and importing that as background - needs to be redone whenever there's a paint event on any of the covered widgets)
>     
>     Also this cannot be resolved by hardcoding some gradient (then matching oxygen, but breaking plastique or qtcurve whch can have any random texture/watermark ...)
> 
> Hugo Pereira Da Costa wrote:
>     Naive question: since QtQuick/QML seems to have limitations, is there a reason for using QML in the compositor kcm, over widgets and ui files, that would not show the issue ? In the past I though using QML inside widgets was discouraged (but maybe this changed, or I missunderstood at that time) ? 
>     
>     Also, I remember AFiestas patching both oxygen and some kde qml elements in order to pass some hints to the style so that things are rendered properly. I have no clue if doable at all (and maybe Alex should be cced here), but could a similar thing be done here to get proper background rendering ? (I can send to the relevant oxygen review requests if needed)
> 
> Hugo Pereira Da Costa wrote:
>     ... sorry. That was actually Martin Klapetek
>     See commit "860401fb393985877573bf136c2d317193da62b9" for instance

> since QtQuick/QML seems to have limitations, is there a reason for using QML in the compositor kcm, over widgets and ui files, that would not show the issue ?

yes sure that's possible, but if we don't use it we'll never find the issues and will never fix them. Thus the situation will always stay the same.

Thus the idea to use Controls in a small KCM to test it is not a bad idea. We still have some months to fix the issues in Qt and in the worst case ship the UI file.


- Martin


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/116648/#review52352
-----------------------------------------------------------


On March 10, 2014, 10:03 a.m., Martin Gräßlin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/116648/
> -----------------------------------------------------------
> 
> (Updated March 10, 2014, 10:03 a.m.)
> 
> 
> Review request for kwin, Plasma and Antonis Tsiapaliokas.
> 
> 
> Repository: kwin-compositing-kcm
> 
> 
> Description
> -------
> 
> Split into one KCM for Desktop Effects and one for Compositing
> 
> Let's try getting the KCM a little bit less scary by properly
> hiding everything the user doesn't have to care about. The prominent
> desktop effects KCM only contains the list of all the effects which
> can be configured and nothing else. Only exception is the disabled
> check after failed GL to make this easier for the user.
> 
> All the "advanced" settings are moved into a new KCM called
> "Compositing" which is put under the hardware component in
> systemsettings. This contains all advanced settings including
> * whether compositing is enabled at all
> * backend
> * animation speeed
> * scale filter
> * unredirect fullscreen
> * color correction
> 
> 
> Diffs
> -----
> 
>   CMakeLists.txt d70672f3937fb44240c2c2d8e2a8d90c67890e2a 
>   kcmkwineffects.desktop PRE-CREATION 
>   kwincompositing.desktop 88ecf46b393df947d6b59ce53c924a82fc65756a 
>   main.cpp 5c68722961be9b8a0a48f3221df7c033ba3cf946 
>   model.h 2efa2a9df63ee7f7ec483df004ff9530fdc0f852 
>   model.cpp 2b25fe04ccacf0fb719956b1b5fa0267611747e6 
>   qml/CompositingView.qml PRE-CREATION 
>   qml/EffectView.qml c1a66dfa41409ba04c4cf4ddded3d95bb4bfaef2 
>   qml/OpenGLErrorView.qml PRE-CREATION 
>   qml/main-compositing.qml PRE-CREATION 
>   qml/main.qml d22dd08da529df938d99ff9d43f52f7b33a7f2d3 
> 
> Diff: https://git.reviewboard.kde.org/r/116648/diff/
> 
> 
> Testing
> -------
> 
> 
> File Attachments
> ----------------
> 
> New advanced compositing KCM
>   https://git.reviewboard.kde.org/media/uploaded/files/2014/03/07/e20bc5a1-c873-497a-b005-efb6b58da7c5__kcmcompositing.png
> New effects KCM
>   https://git.reviewboard.kde.org/media/uploaded/files/2014/03/07/0d6c955e-fdd2-4a84-91b4-53959f2745f5__kcmeffects.png
> Effects KCM with GHNS button
>   https://git.reviewboard.kde.org/media/uploaded/files/2014/03/07/f467b643-5341-4cfe-8282-eeea93bc98e5__kcmeffects.png
> Latest kcmcompositing screenshot
>   https://git.reviewboard.kde.org/media/uploaded/files/2014/03/10/70df036e-f3ad-455d-a323-c8f36220afd4__kcmcompositing.png
> 
> 
> Thanks,
> 
> Martin Gräßlin
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/plasma-devel/attachments/20140310/7b4d24e6/attachment-0001.html>


More information about the Plasma-devel mailing list