Hi,
  <ftp://st.cs.uiuc.edu//Smalltalk/Squeak/2.7/files/Squeak2.8a-beo.zip>
has these updates.  Dan's not kidding folks when he says that things slow
down.  It's amusing for, oh, 15 secs or so, and then it's slow :-)

cheers

bruce

Dan Ingalls <Dan.Ingalls at disney.com> wrote:
> Folks -
> 
> Herewith a number of changes small and mojor.  The major ones install new primitives for LargeIntegers (from Stephan) and for text measuring and display (from Andreas).
> 
> WARNING -- Unless you have a new VM (allowing fast failure from missing primitives), things will slow down by roughly 80%.
> 
> We will try to get a bunch of new VMs out in the next day or so, but meanwhile, you may wish to abstain from updating.
> 
> 	- Dan
> --------------------------
> 2159annotChanged-sw -- Scott Wallace -- 16 May 2000
> When user makes a gesture in the message-list pane of a browser which results in no code selection, the annotation pane is now cleared."
> 
> 2160debugMenu-sw -- Scott Wallace -- 17 May 2000
> A minor improvement to the layout of the debug menu for morphs, plus a new item is added which allows you to inspect a morph's owner chain; and the debug menu now has a 'stay-up' item."
> 
> 2161dAndDConflicts-sw -- Scott Wallace -- 17 May 2000
> Resolves conflicts between update 2157 on the one hand and earlier updates 1843 and 2082 on the other.
> While resolving conflicts involving #invokeBookMenu, also takes the opportunity to improve the menu layout there."
> 
> 2162moreRemovals-sw -- Scott Wallace -- 17 May 2000
> 21 More method removals.  Small potatoes, but potatoes nonetheless"
> 
> 2163LargeIntegersPlugin -- Stephan Rudlof (sr at evolgo.de) -- 9 April 2000
> Special thanks to:	Andrew C. Greenberg
> Speeds up LargeInteger arithmetics.
> Note: If you just want to use the LargeIntegersPlugin as an enduser,
> 	you do not need this change set!
> 	-> Look for a 'LargeIntegers' module (platform specific compiled plugin) and the 'Install_LargeIntegers_module' change set then.
> Preconditions:
> - VM 2.7 or later.
> ToDo:
> File in this change set and generate the module as library according the 'Short Guide for Compiling the LargeIntegersPlugin'.
> Then you will need the corresponding 'Install_LargeIntegers_module' changeset for installing the module.
> If you have generated this module for your platform, because you haven't found it over
> 	http://www.evolgo.com/Squeak/
> please eMail it to (me)
> 	sr at evolgo.de
> with a description of your platform. I will put it onto the corresponding web side then.
> Implementation note:
> There are some extensions of Andrew's TestCodeGenerator and TestInterpreterPlugin in this changeset, Object changes belong to this, too. This could lead to conflicts with other changesets, especially TestCodeGenerator>>initializeCTranslationDictionary  could lead to - easily solvable - problems.
> 
> 2164LargeIntegersInstall -- Stephan Rudlof (sr at evolgo.de) -- 9 April 2000
> Special thanks to:	Andrew C. Greenberg
> This change set binds the Integer classes with the LargeIntegers module (compiled plugin).
> Preconditions:
> - VM 2.8 or later;
> - LargeIntegers module corresponding to your platform (it is called 'LargeIntegers.so' under Linux, for other platforms the name can differ).
> ToDo:
> - *First*: put the LargeIntegers module to the same directory where your 'squeak', '*.image' and '*.changes' are located;
> - Second: file in this change set.
> That's it!
> If something goes wrong you will be notified.
> Functionality:
> It changes the following methods:
> Arithmetic Integer methods
> #digitAdd:, #digitSubtract:,
> #digitMultiply:neg:, #digitDiv:neg:,
> #bitShift:,
> and Large(Positive/Negative)Integer methods #normalize.
> Bit logic Integer methods #bitAnd:, #bitOr: and #bitXor:; they are speeded up, but for positive Integers only!
> All methods are calling corresponding methods in the module now.
> All methods are changed as follows:
> The module will be called automatically, if it is existent.
> If it is not there or fails, the standard ST methods are applied.
> Note: If the module isn't reachable for any reason at startup, the standard ST methods are called automatically. So there is *no* crash risk for installing this change set.
> Note to 2.8:
> Integer changes are permanent: Installing/Deinstalling of the LargeIntegers module will be realized by just putting it in/out the search path for modules.
> 
> 2165dndTransition-sw -- Scott Wallace -- 17 May 2000
> Repairs existing instances of PasteUpMorph and its subclasses so their openness-to-drag-and-drop in the new regime is the same as it was in the old."
> 
> 2166VersionsFix-ar -- Andreas Raab -- 17 May 2000
> Fixes the versions browser to work with the new >16MB changes."
> 
> 2167AnotherInfFormFix -- Andreas Raab -- 17 May 2000
> Fixes just another problem with InfiniteForms."
> 
> 2168SimulatorRevival -- Andreas Raab -- 17 May 2000
> Bring the InterpreterSimulator back to life. A lot has changed with pluginisation so a little effort is required to keep the simulator going. "
> 
> 2169singleStep-sw -- Scott Wallace -- 18 May 2000
> Retards the yellow try-me button in a ScriptEditor so that when you hold it down, it won't fire the script more often than 5 times a second.  This makes it easy to fire a script precisely once by just quickly hitting the try-me button.
> Also, the delay logic associated the 'step' button (in a complex of oval stop-step-go buttons) is improved so that the delay is made a function of how long the step took (formerly, a static 200ms delay had been issued after every step, no matter how long it took to process the step.)" 
> 
> 2170DnDFix-mir
> 
> 2171flexStep-sw -- Scott Wallace -- 18 May 2000
> Fixes a problem revealed by changes in update 2122."
> 
> 2172saveLocalFlaps-sw -- Scott Wallace -- 19 May 2000
> Project-speific flaps are now saved on project save-files.  
> Also, fixes long-standing bug that project save-files would save the flap -- but NOT the tab! -- of open global flaps" 
> 
> 2173platSpecficTweaks-jhm -- John Maloney -- 19 May 2000
> Tweaks to the Win32 and Acorn portions."
> 
> 2174miscFixes-jhm -- John Maloney -- 12 May 2000
> Two unrelated fixes related to running under Netscape:
>   1. fix bug in remote file browsing
>   2. allow exiting interpreter loop"
> 
> 2175OCTweaks-ar -- Andreas Raab -- 22 May 2000
> Two tweaks for OrderedCollection. Makes removals from (large) OCs faster. Make #removeAt: return the removed object."
> 
> 2176MatrixInvert-ar -- Andreas Raab -- 22 May 2000
> Make 4x4 matrix inversion a primitive."
> 
> 2177ColorTweak-ar -- Andreas Raab -- 22 May 2000
> Make Color only return a translucent color if alphaValue is any other than 1.0. Useful for speeding up the computation of pixel ramps in GradientFillStyles."
> 
> 2178PhonemeRecognizer
> 
> 2179CSRework-AA -- Andreas Raab -- 23 May 2000
> Character scanner rework part1. Install the new classes and all associated things without activating them yet." 
> 
> 2180CSRework-AB -- Andreas Raab -- 23 May 2000
> Character scanner rework part2. Activate the new character scanner classes."
> 
> 2181CSRework-AC -- Andreas Raab -- 23 May 2000
> Character scanner rework part3. Reshape StrikeFont and remove all the obsolete character scanner classes."
> 
> 2182CSRework-AD -- Andreas Raab -- 23 May 2000
> Character scanner rework part4. Rename the new character scanner classes and fix up the references to old names."
> 
> 2183FontFix-ar -- Andreas Raab -- 23 May 2000
> Fixes a problem with colored fonts."
> 
> 2184fasterStep-disw -- Dan Ingalls & Scott Wallace -- 23 May 2000
> Changes three kinds of 'run once' buttons, namely the 'step' button and the yellow 'try me' buttons in viewer and scriptor, so that all of them repeat as fast as they can while held down, while preserving the ability reliably to step-exactly-once by simple clicking."
> 
> 2185CrLfFix-mir -- Michael Rueger -- 25 February 2000
> Fixes a bug in CrLfFileStream preventing from correctly detecting lineEndConvention for a crlf file."
> 
> 2186BookMorph-kfr -- Karl Ramberg -- 16 May 2000
> In the rush to make and enhancement I forgot to check for unwanted behavior. Introduced in Update 2096 is the possibility to move a BookMorph by dragging in the PageControls. But the the change ignored the times you don't want the book to be movable. Here is an attempt to fix that.
> Maybe the check should be even more thorough?"
> 
> 2187CelesteEnhFix-mir -- Mike Rutenberg (mdr at scn.org) -- 14 May 2000
> (1) Add full Balloon Help for Celeste buttons and menus
> (2) Minor tweak to updateTOC making huge categories display faster (twice as fast)
> (3) A variety of small fixes including:
> 	* Correct a MailAddressParser table to prevent an infinite loop when evaluating
> 		typed email addresses
> 	* Fix <From filter> and <Subject filter> and <Move Again> button to cancel/fail gracefully"
> 
> 2188BetterFindDup-dvf -- Daniel Vainsencher -- 14 May 2000
> Clean up, speed, and probably fix the findDuplicates. You can definitely delete what this puts in the duplicates category. Emptying the trash before and after is reccomended - it's contents still count."
> 
> 2189Betterbooks-mjg -- Mark Guzdial -- 15 May 2000
> The attached changeset improves the PowerPoint to BookMorph translation process.  This one watches memory, and if you're running low, it writes out all the pages to disk.
> It goes faster if your book has a URL associated with it when you start.  When you create a new book, save the pages to disk, so that it has a URL.  Now just load from first slide of the PowerPoint presentation saved as GIF, and let it go. (And go and go and...it takes awhile :-)"
> 
> 2190CelesteTimeZones-dvf -- Daniel Vainsencher -- 18 May 2000
> sma kindly pointed out that Celeste currently ignores timezone, causing much temporal confusion. Adds a user setting to amend that."
> 
> 2191WatchTweak-bf -- Bert Freudenberg -- 17 May 2000
> Makes WatchMorph a subclass of EllipseMorph, and adds antialiasing. Try a radial gradient from solid to transparent ..."
> 
> 2192EOS-hh -- Helge Horch -- 18 May 2000
> Requires:		Squeak 2.8a
> Modifies ReadStream>>next to signal an Error (EndOfStream) on attempts to read
> at or beyond the end of the stream.  Thanks to Bob Arning (who is silently credited
> in the stamps <g>), existing applications relying on seeing nils should not be
> affected.
> 
> 2193UnixFileDirFix-sr -- Stephan Rudlof -- 8 May 2000
> Description:
> There is a subtle bug under Unix which is fixed by this changeset.
> Possible problem:
> I'm not sure if this is a good manner to fix this problem, or if a deeper refactoring would be better.
> Under Unix '/' is a real directory, and not just a path delimiter like the corresponding characters/strings in other file systems.
> The Unix way expressed in terms of other file systems would be:
> '/' had to be renamed to 'root' and it would be something like 'root/tmp' instead of '/tmp' (this is not a suggestion, just to explain the problem!).
> Other opinions?
> Bug:
> FileDirectory default fileOrDirectoryExists: '/tmp' true
> FileDirectory default fileOrDirectoryExists: '/' false
> After fix:
> FileDirectory default fileOrDirectoryExists: '/tmp' true
> FileDirectory default fileOrDirectoryExists: '/' true
> 
> 2194MIMERefactor1-dvf -- Daniel Vainsencher -- 18 May 2000
> Medium surgery moving some of the text massaging into the MIME classes, and out of it's clients. Makes it trivial to add a stream of stuff to a message as an encoded attachment. 
> Still uses text as the basic medium for changes - direct modification of MIME contents awaits for next refactoring."
> 
> 2195FatBitsPaintFix-raa -- Bob Arning -- 14 May 2000
> Karl: I tryed to edit a 16 bit depth form ( the color palette) to try to make the  - translucent - line on the top of it easier to understand. I explored it, selected it, wrote self bitEdit, and then when I inspect it it has a depth of 32.
> Bob: I see. The 32-bit depth in FatBitsPaint was a change I made since some operations were not working at lower depths. I think you can safely remove the depth test in #accept (included below) and things will work as you expect. I was able to complete your test and see the changes in the ColorPicker."
> 
> 2196DnDClassToggle-len -- Luciano Notarfrancesco -- 17 May 2000
> If the 'class' toggle button is pressed, moving a
> class to other category fails. The attached change
> seems to fix the problem."
> 
> 2197SendingChanges-dvf -- Daniel Vainsencher -- 18 May 2000
> Forget about filing out - use Celeste to directly send changesets across the net, gzipped and all. 
> Warning - might slightly conflict with a recent SqC update (drop history, I think, I'm not up to date).
> Prerequisites -
> MIMRefactor1"
> 
> 2198MoreDiffSup-mas -- Mark Schwenk -- 21 May 2000
> This change set pushes up the diffButton in the optionalButtonRow from MessageList into Browser.
> It also provides diff support for the code pane in ChangeSorters and adds a shiftedMessageMenu item to toggle diff mode.
> So what does this really mean? You can now more easily diff what you see in a code pane against its previous version in more places."
> 
> 2199RomanNumbers-sma -- Stefan Matthias Aust & Andreas Kuckartz -- 21 May 2000
> Andreas wrote: 
> The current method WatchMorph>>createLabels contains the following fragment:
> 	#('I' 'II' 'III' 'IV' 'V' 'VI' 'VII' ' VIII' 'IX' 'X' 'XI' 'XII') at: hour
> Nobody seems to have added a decent method for Roman Numerals to Smalltalk
> in the last twenty years. I wrote one (which was not following the correct
> rules and therefore was not decent) many years ago and posted it in one of
> the Usenet newsgroups. Unfortunately the old Usenet archives at deja.com are
> not accessible for the next few months and I do not have a copy of that
> method...
> Therefore I today have written a new version which hopefully follows the
> rules of the old Romans.
> I rewrote it, reducing its size, named it to #romanString and also added a new method #romanNumber to convert strings back to integers.  --sma"
> 
> 2200BadBadBrowserDnD-mir -- Michael Rueger -- 19 May 2000
> Mark Mayfield wrote:
> > If you have edited some text in the browser and click in the class
> > list or method list you will be prompted if is OK to cancel those
> > changes. If you choose NO...too late...the changes are canceled and
> > you left holding the text of the item you clicked on in the list box.
> Oops, my messing around with Stephans stuff caused this.
> The attached change set should fix this."
> 
> 2201XlucentDND-svp -- Stephen Pair -- 17 May 2000
> Makes the Browser drag and drop appear with white text on a translucent blue
> background.  It also centers the drag morph w.r.t. the cursor."
> 
> 2202WorldBackground-bf -- Bert Freudenberg -- 14 May 2000
> Makes bitmap backgrounds work again (reported by Raymond Asselin). Also translucent colors and gradients work, they're blended with black."
> 
> 2203FileListEnh-ccn -- Chris Norton & Stefan Matthias Aust -- 16 May 2000
> Filing in compressed files results in an error.  So if the user chooses to file in a file with a questionable suffix, I ask the user if he/she wants to inflate the compressed file.  If user says OK, then inflate the file.  If the user says No, then file in the questionable file anyhow.
> sma simplified and rewrote the code, changing it so that after uncompressing the file in continues.  This makes it a really useful feature, IMHO."
> 
> 2204May23Tweaks -- Dan Ingalls -- 23 May 2000
> A couple of fixes for entries in Undeclared.