there is some kind of memory allocation trace in the flex profiler view in Flash Builder. I will try running the app in it. You didn't check in the patch yet, right?
On Sun, May 1, 2011 at 6:02 PM, P T Withington <[email protected]> wrote: > Is there any sort of a leak tool that you can run on teh swf with your test > case? > > I have a bad feeling, because I left your test case open last night and > when I came back my machine was totally hung. I had to force reboot it to > get it back... > > On 2011-04-29, at 18:39, Henry Minsky wrote: > > > Go for it > > > > On Friday, April 29, 2011, P T Withington <[email protected]> wrote: > >> Scratch that. I just tried it. Doesn't improve things that I can see. > >> > >> How about I just check in with the invalidatePixelAligned chopped out? > >> > >> On 2011-04-29, at 18:24, Henry Minsky wrote: > >> > >>> I think that is a good idea, I'll revert it > >>> > >>> On Friday, April 29, 2011, P T Withington <[email protected]> wrote: > >>>> One other thought: > >>>> > >>>> Now that we draw more conservatively, maybe we don't need to pace the > mouse-move events? What if we try reverting r19117? > >>>> > >>>> On 2011-04-29, at 15:25, Henry Minsky wrote: > >>>> > >>>>> Also, subjectively I feel like doubling the frame rate makes it more > >>>>> responsive > >>>>> > >>>>> LFCApplication.stage.frameRate=60 > >>>>> > >>>>> maybe we should make this a default? Or is that an un-neighborly > thing for > >>>>> a downloaded > >>>>> app to do? > >>>>> > >>>>> > >>>>> On Fri, Apr 29, 2011 at 3:16 PM, Henry Minsky < > [email protected]>wrote: > >>>>> > >>>>>> All the calls to invalidatePixelAlignedChildren look like they are > >>>>>> missing > >>>>>> their 'if' clause.... > >>>>>> > >>>>>> public function setY ( newy:Number ):void { > >>>>>> _y = newy; > >>>>>> // Box attributes get scaled > >>>>>> y = newy + ((marginTop + borderTopWidth + paddingTop) * > scaleY); > >>>>>> { invalidatePixelAlignedChildren(); } > >>>>>> > >>>>>> > >>>>>> > >>>>>> On Fri, Apr 29, 2011 at 3:14 PM, Henry Minsky < > [email protected]>wrote: > >>>>>> > >>>>>>> When I stub out the > >>>>>>> > >>>>>>> function invalidatePixelAlignedChildren () { > >>>>>>> return; > >>>>>>> > >>>>>>> then it gets responsive... so maybe that is being run when it does > not > >>>>>>> need to be? > >>>>>>> > >>>>>>> > >>>>>>> On Fri, Apr 29, 2011 at 2:17 PM, Henry Minsky < > [email protected]>wrote: > >>>>>>> > >>>>>>>> Wow that's quite a refactoring! > >>>>>>>> > >>>>>>>> It actually seemed to fix another bug which I hadn't reported yet, > which > >>>>>>>> is in the test case below, the RTE iframe used to get get offset > in the > >>>>>>>> wrong position as you dragged the enclosing window; the further > right you > >>>>>>>> dragged the window, the further the offset. Some bug in computing > >>>>>>>> localtoglobal I think. Anyway it works properly now. > >>>>>>>> > >>>>>>>> > >>>>>>>> However, I see a noticably more sluggish drag behavior now in the > test > >>>>>>>> case below. Don't know if that is the frame rate becoming visible, > or > >>>>>>>> something eating up CPU. Can we now up > >>>>>>>> the Flash frame rate to make up for it? > >>>>>>>> > >>>>>>>> > >>>>>>>> <canvas> > >>>>>>>> <include href="extensions/rte.lzx"/> > >>>>>>>> > >>>>>>>> > >>>>>>>> <stylesheet> > >>>>>>>> boxmodel { > >>>>>>>> padding: 1 3 5 7; > >>>>>>>> border-width: 2 4 6 8; > >>>>>>>> margin: 3 7 11 15; > >>>>>>>> } > >>>>>>>> </stylesheet> > >>>>>>>> <class name="box" extends="view" with="boxmodel" > >>>>>>>> clip="true" x="10%" width="98%" height="50%" > >>>>>>>> shadowblurradius="10" shadowangle="45" shadowdistance="20" > >>>>>>>> shadowcolor="#000000" > >>>>>>>> cornerradius="3 7 11 15" > >>>>>>>> /> > >>>>>>>> > >>>>>>>> <window x="20" y="20" width="500" height="600" resizab > > > > -- > > Henry Minsky > > Software Architect > > [email protected] > > -- Henry Minsky Software Architect [email protected]
