YA-HOO!!

Er, I mean...

OOOOH!  I'm so SCARED!!!

Joshua

On Tue, Oct 31, 2000 at 05:45:35PM -0800, Dan Ingalls wrote:
> Folks -
> 
> I have just released 100 updates worth of ghouls and goblins to strike terror into the heart of your 2.9 alpha images.
> 
> The bulk of the changes have to do with ...
> 	a rewrite of the Morphic Hand and the handling of Morphic events
> 	continued work toward projects that work like swiki pages
> 	reviving and moving SyntaxMorph toward Etoy tile scripts
> 	a start on a movie clip editor
> 
> As stated in the bumper file (which allows the timid to stop updating before all of this), some of the kernel changes may break existing applications.  We have found most of the compatibility problems to be relatively easy to deal with, and we hope you will find the new architecture generally easier to understand and to work with (and we're not done yet either).
> 
> As updates return to a more regular pace, we're hoping soon to put out some further developments in the area of end-user programming for your criticism and enjoyment.
> 
> 	- Dan
> --------------------------
> 2774HTTPSocket-getFix -- Michael Rueger -- 5 October 2000
> Fixes a bug introduced in 2748 preventing http access through proxies. Thanks to Helge for reporting this bug."
> 
> 2775SafetyBumper
> 
> 2776TruthAndBeauty -- Andreas Raab -- 5 October 2000
> The first wave of some new event guts. Installs everything necessary to get us running under the new regime. Cleanup and some other parts will follow shortly."
> 
> 2777HTTPSocket-getFix -- Michael Rueger -- 5 October 2000
> Fixes a bug introduced in 2748 preventing http access through proxies. Thanks to Helge for reporting this bug."
> 
> 2778WeightWatchers-ar -- Andreas Raab -- 5 October 2000
> The Weight Watchers came visiting HandMorph. After they left (only for a brief time) they reported that HandMorph lost:
> - 20 instance variables
> - ~110 methods
> - ~50k in source code.
> So HandMorph just lost half its weight in a single day. Try *that* in the real world ... what a 'Rosskur' (untranslatable)!
> Known pitfall: Some of this has been achieved by moving code around to other parties. But their time will come soon.
> 
> 2779Kimsnav-krRAA -- Kim Rose -- 4 October 2000
> <wanted to change spacing between the rows>"
> 
> 2780ProjStore4-tk
> 
> 2781proxy-fix -- Michael Rueger -- 6 October 2000
> Removes the check for the proxy at Disney Online."
> 
> 2782Flash-segFix -- Michael Rueger -- 6 October 2000
> Workaround for a problem with storing FlashPlayerMorphs in an ImageSegment (storing a sorted collection with it's sort block)."
> 
> 2783DnD-Day -- Andreas Raab -- 7 October 2000
> The third wave of the event rewrite. Fixes tons of issues in the strange protocols involved in drag and drop (although not all). Most of the stuff has been achieved by simply removing all strange looking code and by inserting the most straightforward solution coming to mind (e.g., such as '^true' ;-)
> The ChangeSet also fixes the issue of lagging events and mouse move sends on each cycle. With these changes you will get #mouseMove IF AND ONLY IF the mouse has moved. Mouse move events have a trail of all the intermediate (untransformed) mouse locations recorded with the same state. #mouseEnter/#mouseLeave events are generated by other means (from the hand's mouseOverHandler). ScrollBars have been fixed to take the new regime into account and use stepping rather than relying on implicit sends of #mouseStillDown.
> The issue of parts donors and parts bins has only partly been solved (due to the problem of converting old parts bins). The new regime basically has the following rules:
> * PasteUpMorphs have the hand pick up the top-level object if a mouse down is not handled.
> * When picking up an object which is flagged as a parts donor, the partRepresented (which may not be the morph itself, such as for thumbnails) is being duplicated.
> This restores the basic behavior from before the changes. Other objects which may be flagged as being 'parts donors' and not residing in a paste up need to take care of the appropriate operations itself - as an example PhraseTileMorphs (usually residing in some viewer) handle mouseDown by returning a duplicate of itself if the tile is flagged as being a parts donor.
> Also, the use of a modal shell for pop up menus (introduced during the first wave) has been extended for FillInTheBlanks making those guys really modal.
> 
> 2784DropFixes-ar -- Andreas Raab -- 7 October 2000
> Some fixes related to grabbing/dropping. Removes the last explicit dependencies between hand morph and the tiling system and fixes dropping into flexed environments (verified by playing free cell in an embedded world ;-)"
> 
> 2785HaloTweaks-ar -- Andreas Raab -- 7 October 2000
> Halo tweaks - reverts halo direction and excludes playfields from obtaining the halo automatically unless either it's been clicked on directly or used in an 'inside-out' transition."
> 
> 2786MetaTweaks-ar -- Andreas Raab -- 7 October 2000
> Two more resurrections. Makes menu-driven embedding work again (I forgot to make this work again when putting out wave #3) and makes the control menu appear on the top-most morph within a pasteUp (which I *think* was the behavior we had before)."
> 
> 2787DragsAndClicks-ar -- Andreas Raab -- 7 October 2000
> This change set fixes a deeply buried confusion in #click: vs. #mouseDown: handling. When the event handler was extended to include the #doubleClick: and #startDrag handler the #click: handler was aliased to the #mouseDown: handler, thus completely messing up the notion of #click: being independent from #mouseDown: and only sent in response to some morph actually waiting for clicks.
> While seemingly harmless, the use of #mouseDown instead of an extra #click selector leads to the effect that #click / #drag / #doubleClick: mechanics needed to send #click *whenever* a drag was invoked (all current clients *did* handle both, #mouseDown and #startDrag so #click was essential for restoring accurate behavior). This in turn makes it impossible to distinguish between a drag and a click gesture. While the existing users (such as PluggableLists) did not rely on the explicit distinction between #click: and #drag: (the lists really want to get #mouseDown and are in no way interested in #click:) some of the new mechanics for halos *do* rely on a distinction between a click and a drag (a click being interpreted as transfering the halo and a drag being interpreted as dragging the current halo owner).
> The good news is that because the current users really want #mouseDown and not #click everything will work just fine.
> 
> 2788MoreHaloTweaks-ar -- Andreas Raab -- 7 October 2000
> More halo tweaks. Basically restores the ability to have CMD-drag gesture be different from a CMD-click gesture. CMD-drag will be interpreted as moving the current halo owner, whereas CMD-click is being interpreted as simply putting up a halo on the next halo recipient."
> 
> 2789TwoMoreHalo-ar -- Andreas Raab -- 7 October 2000
> Two more tweaks - shake off halos by a quick click on the desktop and make the halo handles stay inside the world."
> 
> 2790VariousFixes-ar -- Andreas Raab -- 8 October 2000
> Various fixes for broken behavior from the last bunch of changes:
> Halos:
> [X] Transfers of halos through renderers
> [X] Position offset in halo-drags
> [X] Restore all handles after pickup/duplicate
> Grabbing/Dropping
> [X] Automatic repositioning of grabbed morphs (pickup and duplicate handle)
> Scripting
> [X] Touching last tile in Script accidentally drops it as new script.
> [X] TryMe ('!') doesn't work
> [X] Moving and dropping script out of viewer leaves script tile on hand
> FreeCell
> [X] Drag, click, and double click distinction
> 
> 2791MoreCleanup-ar -- Andreas Raab -- 8 October 2000
> Some more cleanup. Removes 15 methods and introduces a single generic version for all of #morphsAt: #morphsAt:addTo: #unlockedMorphsAt: #unlockedMorphsAt:addTo: and their variants.
> 
> 2792StickyFix-ar -- Andreas Raab -- 8 October 2000
> Prevent world menu when clicking on sticky morphs."
> 
> 2793BreakFocus-ar -- Andreas Raab -- 8 October 2000
> Break the current mouse focus when a morph is grabbed or attached by the hand. Put into hand morph because 99% of all clients assume to loose the focus when a hand is about to hold some morph so we'll fix the remaining 1%."
> 
> 2794NebraskaEvents-ar -- Andreas Raab -- 9 October 2000
> Some fixes and encoder methods for events handled from Nebraska."
> 
> 2795browserTweak-raa -- Bob Arning -- 9 October 2000
> - fix overlapping panes in morphic message category browser"
> 
> 2796MenuTweaks-ar -- Andreas Raab -- 9 October 2000
> PopUp menu tweaks. Make leaving a menu item deselect the entry *unless* it's a popUp submenu. In this case defer the deselection by half a second to give the user a chance to reach the submenu before it closes. Timout can be tweaked but half a second seems to be appropriate."
> 
> 2797LastEvent-ar -- Andreas Raab -- 9 October 2000
> Removes the use of 'lastEvent' for determing what a new event should be. Rather buffers the last raw event buffer (which is private to the hand so nobody outside can mess it up) and determines new mouse events based on this.
> NOTE: With the CS we now have 3 different 'lastEventXYZ' instVars in HandMorph. I'll clean this up but for now I'm just trying to get around the image quitting problem (and some other issues which I noticed but never got to the root of the problem). Thanks Bob!
> 
> 2798ConfusedButtons-ar -- Andreas Raab -- 9 October 2000
> The last CS had a little confusion with computing the right set of changed buttons."
> 
> 2799MoreEventFixes -- Andreas Raab -- 10 October 2000
> Tons of fixes all over the place, mostly derived from Scott's and Alan's bug list. Not everything's working yet but the biggest issues have been addressed."
> 
> 2800SevenOfScott -- Andreas Raab -- 10 October 2000
> The next seven items from Scott's bug list. Resistance is futile. All bugs will be assimilated."
> 
> 2801HideSqueakletFolder -- Andreas Raab -- 10 October 2000
> Hide the 'Squeaklets' folder in the default publishing mechanism. It's used as cache only and may be trashed so we better don't allow access to it too freely."
> 
> 2802StickyBooks-ar -- Andreas Raab -- 10 October 2000
> Per BJ+Kims request: If a BookMorph is sticky it should not be picked up when clicking on page controls."
> 
> 2803PolyPainting -- Andreas Raab -- 10 October 2000
> More consistent use of polygons in the sketch editor. The initial mouse-down-drag resizes a triangle; upon mouse up the cursor switches to 'normal' and handles are added. While handles are there, cursor stays normal. Click on handle does 'morphic' poly editing. Click outside will freeze current poly and get back into initial conditions (showing new poly cursor). Tool change will freeze poly."
> 
> 2804UndoFix-ar -- Andreas Raab -- 10 October 2000
> A little fix for undoing a move operation which resulted in a wrong z-order."
> 
> 2805PolyPaintFix -- Andreas Raab -- 10 October 2000
> Fixes a little leftover problem from the poly painting changes which led to creating a new poly when one clicked outside the polygon."
> 
> 2806PublishingFlaps-ar -- Andreas Raab -- 10 October 2000
> Adds the preference #showFlapsWhenPublishing defining whether global flaps and the navigator should be seen when the publisher executes an #armsLengthCommand.
> The default is 'false' leaving us in the illusion that we're still in the same project and prevents that we see multiple overlapping flaps or navigators."
> 
> 2807Remnants-ar -- Andreas Raab -- 10 October 2000
> Some more stuff related to ctrl menu, parts donors and parts bins etc."
> 
> 2808MinorCleanups-ar -- Andreas Raab -- 10 October 2000
> Some minor cleanups after fixing several issues. Mainly removes several #arNotes which are no longer of interest. Also moves #click:/#drag: distinction out of the generic pop up code for halos and puts it specifically into the places where we are interested in drag gestures (which is #blueButtonDown:)."
> 
> 2809DupFix-ar -- Andreas Raab -- 10 October 2000
> Fixes two problems in morph duplication.
> #1: The owner was still set after duplication which leads to strange effects since the duplicated morph should really be ownerless.
> #2: The parts donor property was copied which could lead to other weird effects."
> 
> 2810EventCleanup -- Andreas Raab -- 10 October 2000
> A bit of cleanup in the event stuff. Removes an unused instVar from the mouse over handler and makes sure #handleEvent: is called from all places rather than any of the handler methods directly."
> 
> 2811WnldRevival -- Andreas Raab -- 10 October 2000
> Ah! Finally! One of those great moments where you can throw away one of those incredibly ugly hacks and replace it with something that looks exactly the way it should look. Feels good.
> 
> 2812RetractModalShell -- Andreas Raab -- 10 October 2000
> Retracts the entire damn thing. It was a bad idea."
> 
> 2813MoreTweaks-ar -- Andreas Raab -- 10 October 2000
> Four more tweaks:
> #1: Make #mouseUp behavior consistent with the former behavior in which #mouseUp was only dispatched to the current mouse focus, and make menu items handle them anyways (very nice for popups).
> #2: Make wrap-around behavior of halos more robust (could leave to some spurious jumps otherwise).
> #3: Reactivate some old scroll pane code which I never understood but turns out to be important for the n-pixel inset pane (e.g., clicking *just* on the left or right pixel of a scroll pane leads to strange behavior otherwise).
> #4: Make EventSensor start up with an event queue if we come up running in morphic. We always do these days.
> 
> 2814ShowSqueakletFolder -- Andreas Raab -- 11 October 2000
> Show the Squeaklets folder in the project finder."
> 
> 2815ButtonFixes-ar -- Andreas Raab -- 11 October 2000
> An attempy to make the 'mouse still down' buttons work exactly once on a click. A *quick* click. Also fixes the problem of reissuing some event after a drop operation occured."
> 
> 2816MenuDrag-ar -- Andreas Raab -- 11 October 2000
> Makes a pop-up menu reaquire the mouse focus when dropped at a new location."
> 
> 2817PrimErrorFix-ar -- Andreas Raab -- 11 October 2000
> Make sure the emergency evaluator will act correctly by removing the event queue from Sensor in this very case"
> 
> 2818DnDProp-ar -- Andreas Raab -- 11 October 2000
> The change removes the old #dragNDropEnabled property which was a leftover from several hundred change sets ago..."
> 
> 2819FractionAsFloat-mrm
> 
> 2820psToFileEnh-raa -- Bob Arning -- 12 October 2000
> - allow user to choose file name for DSCPostscriptCanvasToDisk"
> 
> 2821ProjVersions-ar -- Andreas Raab -- 12 October 2000
> Makes project versions be encoded in human readable form, e.g., 
> 	baseName.NNN.ext
> with NNN being the version number."
> 
> 2822TwoMoreFixes -- Andreas Raab -- 12 October 2000
> Two more fixes. Make the control menu pop up for the outermost morph in the innermost world (not the outermost morph in the innermost pasteup) and fix a problem with halos after one of the recent changes. Since mouse events resulting in drops are no longer dispatched separately, the corresponding #mouseUp never reached the duplicate or grab handle after the hand dropped the morph in question. This results in a lonely handle after completing the grab/dup operation."
> 
> 2823FlushCache -- Andreas Raab -- 12 October 2000
> Per Ian's suggestion: Flush the old compiled method when when replacing it."
> 
> 2824MenuItemFix-ar -- Andreas Raab -- 13 October 2000
> A fix for duplicating menu items."
> 
> 2825ProjStore5-tk -- Ted Kaehler -- 11 October 2000
> Better handling of classes that are inside an ImageSegment.  (Uniclasses always go in the segment and are never stored in structures.  If a normal class is in roots, then it goes in the segment and is not in structures.)"
> 
> 2826codeLoader -- Michael Rueger -- 13 October 2000
> Adds compression of code segments and changes the default extension for compressed segments to 'sqz'. This is a workaround for a problem with IE5, which 'intelligently' makes assumptions about 'gz' files.
> Also fixes minor update problems with the progress morph when used from a launcher."
> 
> 2827RWStream-reset -- Michael Rueger -- 13 October 2000
> Fixes a problem with using a RWBinaryOrTextStream for reading objects from internal streams."
> 
> 2828ThreeEventFixes-di -- Dan Ingalls -- 14 October 2000
> Fixes three bugs remaining from the overhaul of HandMorph:
> 	MoviePlayerMorph menu commands don't work
> 	Can't type into CypherPanel
> 	Scrolling seems to get into weird states if you use the elevator
> 
> 2829RWBTStream-reset
> 
> 2830foosCopyFix-sw -- Scott Wallace -- 18 October 2000
> fixes the bug that made inst vars of a foo's copy (in the scripting system) not have the same values as foo."
> 
> 2831enhancedAlphabet-sw -- Scott Wallace -- 18 October 2000
> Adds a couple of special characters to the alphabet holder as per ack request."
> 
> 2832MorphicSteps-ar -- Andreas Raab -- 22 October 2000
> The change set modifies stepping in Morphic in various ways. First of all, the stepList has been reorganized from an OrderedCollection of Arrays to a Heap of StepMessages. Since insertion  and removal are O(log n) operations in Heaps it allows for a very efficient safety feature in step lists. All morphs currently stepped are temporarily taken out of the step list so that an error during step handling will automatically stop the morph from stepping.
> StepMessages also allow a much more general way of handling step messages. The most important part is that a morph can have several different step methods running at once since selector (and optional arguments) can be specified. The main entry for stepping morphs is in Morph>>startStepping:at:arguments:stepTime:. If no explicit step time is given the morph itself will be queried for the #stepTime to allow for dynamic changes of the step frequency.
> Finally, the issue of #mouseStillDown: has been tackled. #mouseStillDown: is now send by step messages with either the last #mouseDown: or #mouseMove: event. A morph wanting to receive #mouseStillDown: needs to return true in #handlesMouseStillDown: and to return a #mouseStillDownStepTime for specifying the frequency for those repeated events. 
> WARNING: Making the frequency very small (e.g., 1ms) makes it IMPOSSIBLE to get exactly one mouse still down event on a reasonably fast machine.
> Known problems: As so often, the event handler aliases (the completely different) notions of #mouseMove: and #mouseStillDown: leading to huge inefficiencies for clients not interested in #mouseStillDown. Sigh. I'm too tired to fix it up right now.
> 
> 2833GrabFix-ar -- Andreas Raab -- 22 October 2000
> A fix for grabbing morphs from transformed environments (such as scroll panes)"
> 
> 2834GrabFix2-ar -- Andreas Raab -- 22 October 2000
> Fixes some side effects from the last CS. Now also makes sure that morphs grabbed from inner worlds come out correctly"
> 
> 2835MethodFinder-tk -- Ted Kaehler -- 19 October 2000
> The examples in the bottom pane of a MethodFinder do not work because the method methodFor: was removed from the system.  This restores it."
> 
> 2836ParseScript1-tk -- Ted Kaehler, Bob Arning -- 23 Oct 2000
> Tiles for any method in the system.  In the message list pane of a Browser, choose 'tile scriptor'.  Bring up a second one to steal parts from.  Dragging a statement removes it from a block.  Dragging anything else drags off a copy. 
> Yellow menu in white space beside method header lets you see the code generated, or accept the method.
> Based on Bob Arning's version 3 of SyntaxMorph."
> 
> 2837RemoteHands-ar -- Andreas Raab -- 24 October 2000
> Removes telemorphic's event handling from HandMorph. RemoteHandMorphs receive their events by registering themselves as event listeners with the hand."
> 
> 2838NebraskaTweaks-ar -- Andi -- 24 October 2000
> Two tweaks for Nebraska:
> #1: Allow remote halos.
> #2: Add a menu item for a '1x1 view'.
> 
> 2839SharedHalos-ar -- Andreas Raab -- 24 October 2000
> Implements the new halo policy for shared environments. The halo is unique to some hand and no interaction of other users will remove the halo automatically. Halos can be obtained by user interaction with the halo."
> 
> 2840RegistryFix-ar -- Andreas Raab -- 24 October 2000
> A preliminary fix for the finalization problem identified by John Macintosh."
> 
> 2841perPlayfieldBatch-sw -- Scott Wallace -- 25 October 2000
> Makes choice of whether or not to 'batch' pen trails be available on a per-playfield basis -- consult the playfield-options menu."
> 
> 2842superSwikiOct-raa -- Bob Arning -- 6 October 2000
> - Preference added to make system windows around project views optional
> - ability provided to publish a project from the project view even if it has never been entered
> - started experiment at swikification of projects
> --- publishing a project now offers a dialog for setting various pieces of information about a project
> --- projects can be stored on a swiki-style server which will retain this additional information and allow searching for projects on same
> --- added ctrl-p to ParagraphEditor to create a text link to another project
> --- publishing a project which does not know its server will use the parent server
> --- new option added to the menu that pops up on the Find button on the ProjectNavigatorMorph 'search the SuperSwiki' which allows searching the swiki for projects matching certain criteria
> --- changed project details to show name at first with button to show all fields
> --- show project details only if name is bad
> --- remove question about whether to create a new project when the url fails - just do it
> --- enabled direct editing of project name in ProjectViewMorph (new ones only)
> 
> 2843ParseScript2-tk -- Ted Kaehler -- 24 October 2000
> Further changes to tile scripts for any Method.  Remove extra SyntaxMorph layer around literals, temps, and braceNodes.
> Put a SelectorNode on the header.
> Made up new nodes for morphs needed for spacing -- KeyWordNode, MessagePartNode."
> 
> 2844Three4Alan -- Andreas Raab -- 25 October 2000
> Three fixes for Alan plus some cleanup.
> #1: Some scripts didn't start stepping. Now they will.
> #2: Grabbing morphs from flexed environments led to strange offsets.
> #3: Dropping flexed morphs in translated environments left the morphs somewhere far outside.
> 
> 2845MouseStillDown-ar -- Andreas Raab -- 25 October 2000
> The change set completely separates #mouseMove from #mouseStillDown by extending the event handler and fixing (hopefully all) affected places. #mouseStillDown is sent after a certain threshold when #mouseDown happened.
> 
> 2846InnerMVCMenus-ar -- Andreas Raab -- 25 October 2000
> Fixes MVC menus running in inner worlds. They would invoke a morphic cycle in an inner world which doesn't work since the event is coming in the reference frame of the outermost world and thus, event processing has to start there."
> 
> 2847NearFinalEventStuff-ar -- Andreas Raab -- 25 October 2000
> Some near-final event stuff. Fixes up the event recorder and provides backward compatibility (e.g., Alan's recorded sessions don't need to be recorded again).
> Also fixes up Nebraska and gets rid of the repeated mouse moves in Nebraska since we don't need them any longer.
> Finally, prepares the conversion methods for the removal of MorphicEvent and MorphicSoundEvent and converts all existing instances.
> 
> 2848FinalHandCleanup -- Andreas Raab -- 26 October 2000
> The final hand cleanup. Removes some more methods an obsolete inst var and fixes some not so nice code. Also removes compatibility code based on #isMorphicEvent. Finally removes the classes MorphicEvent and MorphicSoundEvent"
> 
> 2849EventRename-ar -- Andreas Raab -- 26 October 2000
> Brings the name 'MorphicEvent' back to the game."
> 
> 2850FixFlaps
> 
> 2851projSaveSmall-raa -- Bob Arning -- 26 October 2000
> Some minor fixes (with more than minor impact) for Project saving and loading
> - when saving a project, replace parentProject, previousProject and nextProject with nil so as not to drag extra stuff along. When the project is reloaded, parentProject will be set to the Current project and previous and next will be nil until the project is entered.
> - add class conversion method for NewMorphicEvent (which should only be found in internal projects created while Andreas was tinkering under the hod of events."
> 
> 2852navigatorFix-raa -- Bob Arning -- 26 October 2000
> - restore menu popup ability to SimpleButtonDelayedMenuMorph after recent changes"
> 
> 2853projectStopGaps-raa -- Bob Arning -- 26 October 2000
> Handle the problem of trying to enter a project whose world was turned into a StringMorph by the project save logic. Attempt to load a complete version of the project to replace this stub."
> 
> 2854moreProxies-raa -- Bob Arning -- 26 October 2000
> - ensure proper substitution of DiskProxy for Project when publishing projects that refer to other projects
> - shave a little time off publishing by starting with a bigger segment and not garbage collecting after every failed attempt"
> 
> 2855ownerChangedRevert-di -- Dan Ingalls -- 26 October 2000
> Simply reverts the effect of Update 'LayoutChangedOpt' (internal #2694), which had caused run-around text to fail whenever the text's owner was a PastUpMorph.  A future update, with time for a little testing, will offer a better solution to broadcasting ownerChanged downward when and only when required."
> 
> 2856MoreNebraska-ar -- Andreas Raab -- 26 October 2000
> Removes NebraskaPasteUpMorph and NebraskaWorldState and makes the remote server be associated with the current world."
> 
> 2857FollowThatSqueak -- Andreas Raab -- 26 October 2000
> ... into different projects from Nebraska. The CS transfers any remote connection established with the current world into a newly opened world so all visitors will follow."
> 
> 2858RecursiveFlapFix-ar -- Andreas Raab -- 26 October 2000
> Fixes an infinite recursion in flaps."
> 
> 2859OwnerChanges-ar -- Andreas Raab -- 26 October 2000
> Restores the #fitOnScreen behavior in flap tabs and makes the DropShadowMorph smarter when recomputing its bounds."
> 
> 2860AdaptiveRedraw-ar -- Andreas Raab -- 26 October 2000
> A small tweak with large implications in the etoy system. Makes Thumbnails redrawing themselves based on the time it took to create the thumbnail. For complex objects being viewed most time was spent in computing those thumbnails (for no good reason). The new policy updates the thumbnail after 20 * redraw time attempting to give maximum responsibility."
> 
> 2861DieDropShadow-ar -- Andreas Raab -- 26 October 2000
> Bummer. DropShadowMorph died - it's now incorporated into Morph itself.
> But first transfer all information over to the current wrappees..."
> | hadHalo first |
> DropShadowMorph allInstancesDo:[:ds|
> 	"bullet proof agains obsolete guys"
> 	ds renderedMorph ifNotNil:[
> 		ds renderedMorph setProperty: #hasDropShadow toValue: true.
> 		ds renderedMorph setProperty: #shadowColor toValue: ds color.
> 		ds renderedMorph setProperty: #shadowOffset toValue: ds shadowOffset.
> 		(hadHalo _ ds hasHalo) ifTrue: [ds halo delete].
> 		first _ ds submorphs size > 0 ifTrue: [ds firstSubmorph] ifFalse: [nil].
> 		ds owner ifNotNil: [ds owner addAllMorphs: ds submorphs].
> 		first ifNotNil: [first absorbStateFromRenderer: ds.
> 			hadHalo ifTrue: [first addHalo]].
> 		ds delete].
> 	ds delete].
> 
> 2862superSwikiFix-raa -- Bob Arning -- 27 October 2000
> - small fix for cases where no project info has been entered"
> 
> 2863slimmerPVM-raa -- Bob Arning -- 27 October 2000
> - reduce size of published projects with lots of ProjectViewMorphs. Also fixes some older undesirable behavior"
> 
> 2864sMovie5-di
> 
> 2865ParseScript3-tk -- Ted Kaehler -- 28 October 2000
> Outline in light brown any tile that can be picked up or grabbed.  Outline in green the places where a tile held in the hand may be dropped.  
> Executing 
> 	Preferences enable: #noTileColor
> will suppress tile colors and suppress borders on all but a few kinds of tiles.  This gives a sparse look to a script."
> 
> 2866sMovie5Tweaks-di
> 
> 2867geeWhiz-raa -- Bob Arning -- 29 October 2000
> - restore ability to pick up morphs in a GeeMail with plain click"
> 
> 2868ThreeFixes-ar -- Andreas Raab -- 29 October 2000
> Three fixes for Alan."
> 
> 2869sMovie5DupFix-di -- Dan Ingalls -- 30 October 2000
> Makes green-handle duplication work properly for MoviePlayerMorphs.
> 
> 2870MovieFullSoundtrack
> 
> 2871highGees-raa -- Bob Arning -- 30 October 2000
> - ensure that the pasteup inside the GeeMail is at least as tall as the GeeMail itself"
> 
> 2872DropShadConv-tk -- Ted Kaehler -- 30 October 2000
> Allows DropShadowMorphs in projects written before 2.9(update 2919) to be read in properly.  They are converted to the new idea of being a property of a morph.  The morph tree is sewn together where the DropShadowMorph would have been.
> Improved assistance in writing conversion methods when instances must be converted to a different class."
> 
> 2873B3DBugWorkaround -- Andreas Raab -- 30 October 2000
> Introduces an image workaround for a very bad bug within the 3D engine. The funny part about this particular bug is that it did almost never occur before - it only happens if the rendering process gets interrupted (e.g., through a process switch) after executing one particular allocation and when being used with quads. With the recent changes in the event sensor these chances have dramatically increased...
> Oh, yeah, and it also fixes the actual bug."
> 
> 2874FlapFix-ar -- Andreas Raab -- 30 October 2000
> Fixes disembodied flap tabs."
> 
> 2875Conversions-ar -- Andreas Raab -- 30 October 2000
> Two more conversion methods"
> 
>