- User Since
- Apr 17 2015, 10:32 PM (169 w, 2 d)
Sat, Jul 14
If this is an issue, we could try looking into setting a Q_DECLARE_TYPEINFO on the WindowQuadList.
Right now QList still allocates a second set of memory for complex types separate from the internal void* list that we're pre-allocating here.
Fri, Jul 13
And in that function qreal newScale = screens()->scale(m_screen); gets the value from here: https://github.com/KDE/kwin/blob/master/screens.cpp#L114
You can see that this function just returns 1
One can shield
shielding means we've already gone wrong, and you'll still get broken shadows on the first tabbox.
It (currently) depends. There are two scale UI options with the appropriate one visible.
Would have expected some fancy std::remove_if from you
Thu, Jul 12
Sorry for the delay
I think micro-optimizing the tests further is simply not worth anyone's time
Wed, Jul 11
Done some investigation.
Needs a comment in the code explaining why.
Is this now redundant when this KCM was remodelled ?
Tue, Jul 10
this has been fixed in QtQuick.Controls 2.0
Sat, Jul 7
My fault :/
Thanks for taking care of that.
Fri, Jul 6
This may not be the window you started with! You need to track which window you started with, if it got destroyed you need to ask the focuschain
Meh I still don't understand why KPluginInfo(KService) is deprecated it makes no sense, given using KService for plugins is not yet deprecated.
Actually increases ram usage of sddm-greeter by 10 MiB here. I thought it would never load the image larger than it is? :/
Please open a bug report that this is missing and should be ported.
I think we should focus on getting the core done here, and we can do the effects themselves in a different patch.
org.freedesktop.Application (KDBusService ) supports a variant map of "platform-hints" X startup ID is currently in there and then not used.
Thu, Jul 5
Yeah, all the effects will need some tweaks for the cardinality changes.
Unfortunately there are code areas where the way to set it is through the Plasmashell API and other areas where it is through the internal setting mechanism.
What's the status of this?
Wed, Jul 4
Good analysis, that's what I like to see at the start of any changes!
What's the state of an xwayland client being on N desktops?
Or would it make sense to wait on merging until Qt 5.12 with XDG WM Base support is available to have more test candidates?
From IRC today.
Tue, Jul 3
In an email
Mon, Jul 2
The parental changes totally make sense. +1 to that.
For scripted effects, yes, they're packages
For a 3rd party C++ plugin effects with a matching .desktop file and this will no longer have the right data.
This doesn't seem right. You're still using a mixture of KService and KPluginInfo just avoiding the warning by parsing everything twice, but not really the point of the warning.
In another patch I'm doing a V2 of this interface in the next frameworks cycle.
May as well try to merge this at the same time so we only have 1 version bump?
Q_PROPERTY(int foo MEMBER m_foo) saves writing getters just for moc.
Not exactly a hotspot.
Sun, Jul 1
Const find/ constEnd should work here.
Sat, Jun 30