Hi,

Thanks, this is now available at:

ftp://st.cs.uiuc.edu/Smalltalk/Squeak/3.3alpha/Squeak3.3a-4981.zip.

cheers

bruce

Scott Wallace <scott.wallace at squeakland.org> wrote:
> Twenty new updates are now available on the external 3.3a update 
> servers.  These are mostly bug-fixes.
> 
>    -- Scott
> 
> 
> 4962BBFixAlphaRule -- Andreas Raab -- 21 August 2002
> Adds a #fixAlpha:with: combination rule to BitBlt.
> 
> 4963searchTweaks-sw -- Scott Wallace -- 22 August 2002
> Various tweaks and enhancements to the search-in-viewer facility, including...
> * Makes the search-in-viewer work when the search button is pressed 
> as well when enter or return is used to submit the search.
> * When a user submits a search request in a Viewer and no hits are 
> found, explicitly inform the user about this.
> * Adds to the menu for every phrase in a viewer the item 'show 
> categories...', which will present you with a list of all the viewer 
> categories in which the item will be found; if you choose any of 
> these, a viewer pane specifically for that category will replace the 
> searching-viewer pane
> 
> 4964debuggerButtons-sw -- Scott Wallace -- 22 August 2002
> Makes the conventional code-tool buttons (senders, implementors, 
> versions, inst vars, etc.) available in a debugger, if the 
> #optionalButtons preference is true.  This is independent of the 
> debugger-specific button pane (Proceed, Restart, Send, etc.), which 
> is always shown and was never governed by the optionalButtons 
> preference anyway.
> This change does not affect mvc debuggers.
> 
> 4965moreViewerTweaks-sw -- Scott Wallace -- 23 August 2002
> Further tweaks to search-panes-in-viewer:
> * Make the wording of the help-message for the search button be 
> couched in terms of tiles, not selectors.
> * Add new search panes, when they are requested, at the top of the 
> viewer rather than at the bottom.
> 
> 4966fromAppleHill-sw -- Scott Wallace -- 4 September 2002
> Three issues that arose in the company of Kim and BJ at Apple Hill:
> (1)  Newly drawn sketches sometimes did not show up in the playerlist 
> of the Presenter (because its playerList cache was not getting 
> invalidated;) thus their names would not show up in the script-name 
> popup, and also sometimes they would not get stepped.  This is a bug 
> that actually goes back more than four years.
> (2)  Launching a PDAMorph in a system that lacked a .sources file, 
> (such as a typical PlugIn image), would raise an error, because 
> PDA.initialize assumed that class comments would always respond to 
> the selector #string, which Texts do but which (perhaps surprisingly) 
> Strings themselves do not.
> (3)  For the Squeakland image, make four custom preference adjustments.
> This update is intended to apply both to 3.3a and the Squeakland 
> plug-in update streams; the preference changes will only be felt in 
> Squeakland
> 
> 4967B3DRendererFlags -- Andreas Raab -- 5 September 2002
> Adds renderer creation flags for creating HWA renderers. Besides 
> using software/hardware renderers (which are now handled through the 
> flags and no longer through individual booleans) a new flag is 
> introduced to indicate the request for stencil buffering."
> B3DHardwareEngine addClassVarName:'B3DHardwareRenderer'.
> B3DHardwareEngine addClassVarName:'B3DSoftwareRenderer'.
> B3DHardwareEngine addClassVarName:'B3DStencilBuffer'.
> 
> 4968FixBalloonModule -- Andreas Raab -- 5 September 2002
> (Incompletely) fix the location of BalloonEngineConstants. Somehow it 
> got moved from Balloon.Engine to Balloon.Drawing although there is no 
> trace in the updates I can see. The CS at least allows generation of 
> new VMs but it is incomplete since there are various other references 
> that are NOT fixed.
> 
> 4969divers-sw -- Scott Wallace -- 4 September 2002
> * Fixes the bug that would drop you into a debugger when you used the 
> 'find window' feature when there was a window with a zero-length 
> title.
> * Only show the extra row of buttons in the debugger (as per update 
> 4964) if the preference #extraDebuggerButtons  is true. To minimize 
> upset.
> * When a message-list shows a class definition or hierarchy, now 
> shows something appropriate in the annotation pane.
> * Repairs the long-standing annoyance that methods imported via 
> change-list browsers and versions browsers did not get noted in the 
> recent-submissions list. and hence would not be seen in 
> recent-submissions browsers.
> * Fixes a bug that could make a pop-out behavior change for flap tabs 
> not work properly.
> * Fixes the bug that String.withFirsCharacterDownshifted would 
> generate an error when called with an empty string.
> 
> 4970overlaps-sw -- Scott Wallace, Leo Burd, John Maloney -- 6 September 2002
> Adds an 'overlaps' test to the etoy vocabulary.  This is simple 
> object-to-object collision detection, and is smart about jagged 
> edges, transparency, etc.
> NOTE:  The 'touchesA:' test is changed to exploit this better 
> collision detection (previously it had only looked for bounding-box 
> intersection) but is otherwise left with its quite inscrutable 
> semantics (consult comment in Player>>touchesA: for details.)
> Also, this update changes the behavior of the 'show categories' 
> function from the viewer-entry menu; now, instead of replacing an 
> existing viewer pane with one pointing at the new category, it *adds* 
> a new viewer pane showing the new category -- per Kim's suggestion
> 
> 4971fasterListOfMany-dvf -- Daniel Vainsencher -- 7 July 2001 (originally ;-))
> This update prevents significant slowdown for Squeak when looking at 
> really long lists that use this morph (happened to me using the 
> ""recently logged changes"" feature on a changes file with lots of 
> updates)
> drawOn: for this class was O(N*N) in the number of list items, now O(N).
> sw 8/30/2002: merged Daniel's speedup with two other changes to this 
> method that have appeared in the update stream in the meantime"
> 
> 4972fixWorldGradient-ar -- Andreas Raab -- 7 September 2002
> Fixes the problem when trying to set the world's gradient color
> 
> 4973xmlFix-mlr -- Michael Rueger -- 9 September 2002
> Convert to uppercase before parsing numbers, in order properly to 
> support bases > 10.
> Replace 16-bit characters with spaces.
> 
> 4974knownRenames-mlr
> 
> 4975More3DPrims-ar -- Andreas Raab -- 18 September 2002
> Make four more matrix operations run in primitives for speed and less 
> pressure on GC.
> 
> 4976B3DPrimFix -- Andreas Raab -- 27 September 2002
> Fixes a bad mistake in one of the new primitives.
> 
> 4977fixesEtc-sw -- Scott Wallace -- 3 October 2002
> * Fixes 'one last' bug in the revamped implementation of #spawn: if 
> the spawning browser were looking at the #all category, the spawned 
> browser was being given a faulty message-category-list index, which 
> would result in difficulty submitting the method later on.
> * Fixes the annotation for class-comment for the case when the 
> selection is a metaclass method.
> * Fixes Wonderland.makePlaneNamed so that it returns the resulting actor.
> * Fixes Scamper.back so that it does not generate an error when the 
> document index is out of range.
> * Bulletproofs MenuItemMorph.mouseEnter against an orphaned condition 
> that is known to occur.
> * Fixes bug that had made text on menu tiles editable; click on menu 
> tile again brings up the list of alternatives.  I don't know just 
> when or why this got broken, but it was reported recently by Alex 
> Peake.
> * Fixes MenuItemMorph.veryDeepFixupWith: so that it does not bomb 
> when #arguments is nil.
> * Adds a menu item for Recent Submissions browsers that allows the 
> currently-selected method to be removed from the RecentSubmissions 
> list.  This is useful to rid the Recent Submissions browser of 
> methods that were temporarily changed for debugging and subsequently 
> reverted.
> * FileList2 will now return a read-only stream on a file if the 
> default read/write-access cannot be obtained.  Practically, speaking, 
> all known uses of FileList actually only need read-only access anyway.
> 
> 4978lengthEtc-sw -- Scott Wallace -- 3 October 2002
> * In viewers, replaces the former 'height' and 'width' variables with 
> 'length' and a redefined 'width'.
> * Length is measured in the direction the object is heading,  Width 
> is the orthogonal measure.  These two measures are now invariant as 
> an object rotates (except for small round-off artifacts.)
> * Adds (this is a temporary name, until Alan tells me what the real 
> name is ;-) #headingTheta, which gives the angle from positive-x that 
> an object's major axis (i.e. the direction in which the 'length' is 
> computed) makes.
> * Puts commands for switching into and out of 'vector mode' into the 
> personal menu for Alan (i.e. only if the hard-coded #isFlagship 
> preference is true.)
> * Note that the length and width computations for PolygonMorphs are 
> very tortured and special-cased and slow.  I would welcome this being 
> cleaned up.  AFAIK it is only PolygonMorph that is problematical in 
> this way.
> * For the particular forcing function behind this work, namely Alan's 
> 'players-as-vectors' etoy work, PolygonMorphs that have precisely two 
> vertices are special-cased to have intuitively correctly behavior for 
> length and width; the Arrow found in the Objects catalog is a good 
> prototype to use.
> * Fixes a long-standing bug in 
> methodInterfacesInPresentationOrderFrom:forCategory:
> 
> 4979decimalPlaces-sw -- Scott Wallace -- 3 October 2002
> * When the user text-edits the value in an UpdatingStringMorph that 
> holds a numeric value (e.g. in a Viewer or Watcher,) the user's 
> type-in is now parsed to determine how many decimal places to show in 
> that readout from then on.  Thanks to Kim Rose for this suggestion.
> * When an UpdatingStringMorph has a particular number of 
> decimal-places specified, that number of decimal places is now always 
> actually shown.  Formerly integers were shown without decimal points 
> completely, whatever the number of decimal places specified, and 
> trailing zeros were generally not shown.  Now, if you have asked for 
> two decimal places, the readout will always show precisely two digits 
> after the decimal point.
> * When the user asserts a particular number of decimal places in a 
> readout for an etoy built-in variable such as x or width, that 
> preference is remembered and used whenever that variable is 
> subsequently viewed.
> * When an arrow-setter phrase is torn off from a viewer, if it 
> pertains to a numeric variable for which a decimal-places preference 
> has been expressed, that preference is reflected on the constant tile 
> in the spawned assignment phrase.
> * The #heading pseudo-slot is no longer gratuitously rounded.
> * The main viewer-entry menu (activated by clicking on the menu icon 
> next to a variable name in a viewer) now contains a 
> set-decimal-places item for *all* numeric-valued variables, not just 
> for user-defined instance variables.
> 
> 4980thumbHolderFix-bf -- Bert Freudenberg -- 30 September 2002
> Makes getValueAtCursor and getGraphicAtCursor for a Holder showing 
> thumbnails return the represented morph. Before, the thumbnail itself 
> was returned which caused the animation to be played at thumbnail 
> size.
> 
> 4981alphaDropTest-ar -- Andreas Raab -- 17 September 2002
> Fixes a weird problem with translucent text drop shadows.