Hi henrik
I added an issue
http://code.google.com/p/pharo/issues/detail?id=629
so that we did not forget it
Stef
On Feb 25, 2009, at 2:36 PM, Henrik Johansen wrote:
> Had a look at drawWorld:submorphs:invalidAreasOn: today, the
> "Experimental top-down drawing" there does not seem to be working
> optimally.
> Basically, when there's many windows behind one you're resizing,
> this many times leads to ALL of them getting repainted.
> Attached is a fix for this, so only the system window areas which
> are visible and affected are updated.
> It will not speed up cases when resizing a complex morph (like the
> Changes Log), but helps when resizing simple ones with many other
> windows in the background.
> If anyone interested in faster resizing (without fastDragWindow)
> would review the code/check if it helps them it'd be nice.
> A simple test is to open up a couple system browsers, put them on
> top of eachother, and resize a simple window (Like the "Welcome to
> Pharo" window) over their bounds.
> Then load the changes, and see if you notice a difference.
> In a test i did (same windows open/resized, as described above),
> time spent in drawWorld:submorphs:invalidAreasOn: was reduced from
> 86% to 65.7% after filing in changes.
>
> If anyone with a better understanding of Morphic know if anything
> larger than the clippingBounds is safe to use as default in
> Morph>>areasRemainingToFill: , that would probably help as well.
>
> Cheers,
> Henry
> 'From Pharo0.1 of 16 May 2008 [Latest update: #10243] on 25 February
> 2009 at 2:30:25 pm'!
>
> !Morph methodsFor: 'drawing' stamp: 'Henrik Sperre Johansen
> 2/25/2009 12:06'!
> areasRemainingToFill: aRectangle
> "May be overridden by any subclasses with opaque regions"
>
> ^ Array withAll: (aRectangle areasOutside: self clippingBounds)! !
>
>
> !WorldState methodsFor: 'update cycle' stamp: 'Henrik Sperre
> Johansen 2/25/2009 13:52'!
> drawWorld: aWorld submorphs: submorphs invalidAreasOn: aCanvas
> "Redraw the damaged areas of the given canvas and clear the damage
> list. Return a collection of the areas that
> were redrawn."
>
> | rectList c i n mm morphs rects rectToFill remnants
> remnantIntersects rect validList |
> rectList := damageRecorder invalidRectsFullBounds: aWorld viewBox.
> "sort by areas to draw largest portions first"
> rectList := rectList asArray sort: [:r1 :r2 | r1 area > r2 area].
> damageRecorder reset.
> n := submorphs size.
> morphs := OrderedCollection new: n * 2.
> rects := OrderedCollection new: n * 2.
> validList := OrderedCollection new: n * 2.
> rectList do:
> [:dirtyRect |
> dirtyRect allAreasOutsideList: validList
> do:
> [:r |
> "Experimental top-down drawing --
> Traverses top to bottom, stopping if the entire area is
> filled.
> If only a single rectangle remains, then continue with
> the
> reduced rectangle."
>
> rectToFill := r.
> remnants := OrderedCollection with: r.
> i := 1.
> [remnants isEmpty or: [i > n]]
> whileFalse:
> [mm := submorphs at: i.
> ((remnantIntersects :=
> remnants select: [:each | (mm
> fullBounds intersects: each)]) notEmpty and: [mm visible])
> ifTrue:
> [morphs
> addLast: mm.
>
> rects
> addLast: (Rectangle merging: (remnantIntersects
> collect: [:each | mm fullBounds intersect: each])).
>
> remnants removeAll: remnantIntersects.
>
> remnantIntersects do: [:eachIntersect | remnants addAll:
> (mm areasRemainingToFill: eachIntersect)].
>
> remnants size = 1 ifTrue: [rectToFill := remnants first].
>
> remnants isEmpty ifTrue: [rectToFill := nil]].
> i := i + 1].
> "Now paint from bottom to top, but
> using the reduced rectangles."
> rectToFill
> ifNotNil: [aWorld drawOn: (c :=
> aCanvas copyClipRect:
> rectToFill)].
> [morphs isEmpty] whileFalse:
> [(rect := rects
> removeLast) == rectToFill
> ifFalse: [c :=
> aCanvas copyClipRect: (rectToFill := rect)].
> c fullDrawMorph: morphs
> removeLast].
> morphs reset.
> rects reset.
> validList add: r]].
> ^validList! !
>
> _______________________________________________
> Pharo-project mailing list
> [email protected]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________
Pharo-project mailing list
[email protected]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
