Evolving Appdash

[Eike Hein]

Some relevant IRC log:

[15:37] <Sho_> mgraesslin: i wonder if we can get a kwin effect that
only blurs in the desktop layer and cuts out the windows, as a fallback
or even default for dashboard
[15:37] <Sho_> mgraesslin: or we can resurrect the kde3 approach to
translucency and have plasma export the wallpaper and just blur that in
[15:38] <mgraesslin> Sho_: then I need to ask why kwin effect at all,
Plasma has access to the wallpaper
[15:38] <Sho_> in theory this doesn't solve much because we support
animated wallpapers, but in the common case ...
[15:38] <Sho_> mgraesslin: yeah
[15:38] <mgraesslin> Sho_: you could just use the "fake blurred"
wallpaper from lockscreen
[15:38] <notmart> hmm, don't know whatthis would save..
[15:39] <notmart> would just look rather broken.. a solid color would
already be better
[15:39] <Sho_> notmart: mgraesslin is concerned about performance of the
blur effect on bad drivers/gpus when there are changing windows below
the dashboard
[15:39] <notmart> or no blur at all just 90-95% opacity
[15:39] <notmart> yeah, i know
[15:39] <Sho_> i'm not sure it would look broken, because
[15:39] <Sho_> - there's actually a user wishlist item for this
[15:39] <mgraesslin> Sho_: the problem is not the performance, the
problem is that the effect is not meant for that
[15:39] <notmart> well, the kde3 effect sure looked very broken ;)
[15:39] <Sho_> - windows added the same option to metro in 8.1 on user
request (over solid colors previously)
[15:40] <mgraesslin> Sho_: and blur just messes up the rendering, we
need to render to FBO, read back, apply shader on it
[15:40] <mgraesslin> Sho_: also means no chance to move it to DRM layers
[15:41] <bshah> mgraesslin: log sounds sane?
[15:41] <mgraesslin> bshah: just pushed a new variant
http://commits.kde.org/scratch/graesslin/plasma-integration/abb1492527f1b8d77d2b23faba4eff2af64a3543
[15:41] <mgraesslin> bshah: now with --prune-empty
[15:41] <-- _rd_ (~rd at p5B2C73DB.dip0.t-ipconnect.de) has left this
server (Ping timeout: 240 seconds).
[15:42] <bshah> mgraesslin: dunno but this doesn't sound more cleaner
[15:42] <mgraesslin> bshah: it is, consider
https://quickgit.kde.org/?p=scratch%2Fgraesslin%2Fplasma-integration.git&a=commitdiff&h=d2eecfddce21356fb2d5ec7afa1b5ab8dd6bdc89
[15:42] <bshah> I would first put this platform theme in seperate folder
in frameworkintegration
[15:43] <mgraesslin> bshah: it already is
[15:43] <mgraesslin> bshah: I just renamed it
[15:44] <notmart> well, there must be a way to do blur fast enough, not
supporting a feature that is universal everywhere else doesn't sound
acceptable to me, even if is for speed reasons
[15:45] <-- toscalix (~agustinbe at 149.100.1.63) has left this server
(Quit: Konversation terminated!).
[15:45] <mgraesslin> notmart: sure it can be done fast, but not with the
shader we have
[15:45] <d_ed> I can make it 4 times faster in a single word.
[15:45] <mgraesslin> notmart: our logout effect has a fast blur
[15:45] --> toscalix (~agustinbe at 149.100.1.63) has joined this channel.
[15:46] <d_ed> downsample - you're blurring anyway, you won't notice
[15:46] <notmart> mgraesslin: soo, can we change it to use that one?
[15:46] <mgraesslin> d_ed: yeah, that's what logout does
[15:47] <notmart> or are there reasons why dialog has to use that one?
[15:47] <mgraesslin> notmart: well it needs a new effect, that's the
whole point. The generic blur doesn't work for that
[15:47] <notmart> couldn't the dashboard use the exact same one as the
logout effect?
[15:47] <mgraesslin> notmart: no, because the logout works completely
different and does the vignetting and all that
[15:49] <notmart> but should be easy enough to modify it on a new
effect, no?
[15:49] <mgraesslin> yes
[15:49] <-- crog_ (crog at nat/redhat/x-bkuevcoegobxtsld) has left this
server (Remote host closed the connection).
[15:49] <notmart> remove the vignetting, call it fullscreenblur, profit
[15:49] <mgraesslin> notmart: but we need to be careful, appdash is not
fullscreen
[15:50] <mgraesslin> it's only filling a screen
[15:50] <mgraesslin> which means on a multi-screen system only one
screen is covered
[15:50] <mgraesslin> which affects the rendering algorithm