Squeak-dev September 2000

squeak-dev@lists.squeakfoundation.org

115 participants
411 discussions

Re: Browser plugin (was: Re: Mac 2.8 Browser plugin)
by Bob Arning 01 Sep '00

01 Sep '00

On Thu, 31 Aug 2000 23:49:21 -0700 Michael Rueger <m.rueger(a)acm.org> wrote: >> >* You shouldn't use HTTPSocket to get stuff, but >> > (FileStream requestURLStream: 'squeak.3ds') >> > which asks Netscape to fetch the file. You might also look at the >> > PluginCodeLoader and PluginHTTPRequest stuff. >> >> Bert, >> >> Wouldn't this prevent those squeaklets from being used in Squeak that is *not* in a browser? > >No, the code falls back to the HTTPSocket get when Squeak is not running >inside a browser. Or it will as soon as 'FileStream requestURLStream:' exists in a normal image, right? ;-) Cheers, Bob

1 0

Re: A Squeak Speaklet
by Karl Ramberg 01 Sep '00

01 Sep '00

Bert Freudenberg wrote: > On Fri, 1 Sep 2000 Torsten.Bergmann(a)phaidros.com wrote: > > > It works here, but I didnt use Berts Plugin image. I've produced one > > of my own. Some tips: > > [...] > > Nice description, Torsten. > > The only problem is that you better should start with a virgin 2.8 image. > This is the current release, which we can expect others to have. If you > use 2.9a, you risc using features that are not yet in 2.8. > > -- Bert Shouldn't the plugin check the image version to notify you need and update ? This is official updates ofcourse like Smalltalk version == 'Squeak 2.8' ifFalse:[ get the new image form Berts server]

2 2

[Bug] Re: [updates] another 5 for 2.9alpha
by Karl Ramberg 01 Sep '00

01 Sep '00

Dan Ingalls wrote: > 2551SelectionTweaks-di -- Dan Ingalls -- 31 August 2000 > Fixes a bug that caused shift-drag of polygon vertices to grow the vertex handles. > Limits group selection to the submorphs of the pasteup in which the selection was begun. > Allows a group selection to be augmented or depleted by secondary selections. > Elides close curve options for lines with only two vertices. Selection rectangle does not delete it self. Karl

1 0

[Updates] 22 new ones
by Dan Ingalls 01 Sep '00

01 Sep '00

Folks - I have just posted 22 new updates, which are summarized here below. Enjoy --------------------------- 2525EToyComm3-raa -- Bob Arning -- 18 August 2000 - fixed the 'send on return' option in multi-chat - made cachedObjects an instance var instead of class var in CanvasEncoder to better support multiple clients. There is probably still room for improvement in the handling multiple clients. --some actual code---" CanvasEncoder stopCaching. 2526paintBox4-tk -- Ted Kaehler -- 22 August 2000 Changes to the PaintBox: When you start painting, the PaintBox will do its best to pick a position fully within the current world. All flop-out parts will appear in bounds. (PaintBox may sometimes appear on top of the painting. If you accidentally make that happen, tell me how.) Both the 'shapes' and 'stamps' drawers can show at once, as Kim asked. ColorPicker and PaintBox both come to the front when used. (Could go behind painting before) The PaintBox itself has a new item on its menu (via the red handle). 'Grab stamp from screen' allows you to grab any section of the screen and make it a stamp. Very useful for combining two images into one sketch. A normal stamp can't do this because it leaves pickup mode when it leaves the paint area. When you want to make a stamp from within the picture, use the normal stamp icon on the blank stamp. 2527tileLayout-tk -- Ted Kaehler -- 28 August 2000 A previous update made EToy tiles slightly narrower than they were before. Existing scripts did not know to re-lay themselves out. This tells them to do it. Change color of Fancy Watchers and make them not as tall." 2528specialDrag-sw -- Scott Wallace -- 28 August 2000 Changes the behavior of cmd-drag so that in no circumstances is a mouseUp: issued to the draggee at the end of the drag. This now generically circumvents a long-standing problem that various morphs have had to work around in various ad-hoc ways. Those old workarounds are not however removed by this fileout, though someone should do that sometime." 2529polygonMenus-sw -- Scott Wallace -- 28 August 2000 Makes the wording and availability of menu items relating to polygons be more dynamically responsive to changing circumstances, thus allowing persistent menus to remain correct as polygon properties change." 2530paintingGoof-raa -- Bob Arning -- 29 August 2000 - fixed a few problems occurring after use of scale or rotate buttons" 2531InfiniteNebraska-raa -- Bob Arning -- 28 August 2000 - support for drawing InfiniteForms through Nebraska - reduced memory requirements on client side" 2532miscellany-sw -- Scott Wallace -- 29 August 2000 * A workaround for a recently-introduced bug in Morph.submorphNamed:ifNone:. * Don't step the models of collapsed system windows. * The programmatic compilations of Preferences methods no longer clutter up the current change set. * Performance improvements relating to computationof referencePlayfield" 2533miscellanyFixup-sw -- Scott Wallace -- 29 August 2000 Reincorporates code omitted from the recent miscellany fileout because of a mechanical error: * Fixed bug in ChangedMessageSet -- was not getting annotation panes invalidated properly. * Adds #selectAll to ParagraphEditor" 2534undo-sw -- Scott Wallace -- 29 August 2000 Provides a simple foundation for implementation of UNDO in morphic. * If you have an action that you want to make undoable, you need to create a Command object, and arm it so that it preserves state needed for undo/redo, and give it blocks for carrying out undo and redo (see HandMorph.grabMorphInner and ColorPickerMorph.selector: for two examples of how to do this -- the first for undoing morph-move commands, the second for undoing a morph recolor command.) * When the command is to be *done*, you tell the system to install it as the current command; this is done by sending the message Command rememberCommand: aCommand * If you field a user request that makes the last command no longer undoable, you can send the message Command cannotUndo. In this initial burst of Undo implementation, full undo is provided for: - All color changes, including: Fill-color Border color Text-highlight color Insertion-point color Second gradient color - Move - Resize - Rotate An architecture is also in place for 'infiniteUndo', but the preference governing this is currently hard-coded to false, pending further work. A mechanism is added to intercept unfocusedKeystrokes, i.e. keystrokes typed when no morph clalims the focus. Used, for example, for cmd-z implementation for undo, as well as for a tab-among-fields feature." 2535fastColorChange-sw -- Scott Wallace -- 29 August 2000 Drastically reduces the number of steps required to change a color in morphic. Adds a halo item on recolorable morphs that gives you a color picker right away. Adds a preference that allow you to specify the use of modal color pickers. A modal picker goes away once you have made your choice; and until you've made your choice, it keeps control of the mouse. It shows an eyedropper cursor. When the modal picker is present, it continuously reports (in its little feedback box) the color under the mouse, waiting for the mouse to go down. Once the mouse goes down, and as long as it stays down, the color under the mouse is applied to the target continuously. The color under the mouse when the button comes back up will be the color selected, with the degree of transparency prevailing at the time." 2536MorphicDraw-di -- Dan Ingalls -- 29 August 2000 Sundry changes to make Morphic a more useful environment for graphical ccomposition. Ability to make synthetic bold for fonts with chars that have 0 leading. Introduces LineMorph as a macro for degenerate polygons. Introduces shift-click for showing and hiding handles of curves and polygons Shift-click near a hidden vertex shows vertices, and starts drag, then hides vertices. Introduces a menu item to mutate a polygon to a curve and back again. Fixes resize so that it no longer clips a pixel when extending the bounds. Introduces a selection morph for deletion, moving, duplication, color change, etc. Use by dragging in an empty place in the world, or choose 'make a selection' from a PasteUpMorph's menu. Replicate feature: duplicate/move/duplicate of a selection repeats the same spacing. [Note: the changed order of #layoutChanged and #changed in Morph>>extent: and Morph>>position: is fundamental, and could cause strange behavior almost anywhere. It should only be an improvement -- it fixes a long-standing bug that would clip pixels while extending morphs with either the resize command or by the halo -- but it could break dependent code]. 2537BetterGridding-di -- Dan Ingalls -- 29 August 2000 Gridding is now specified in PasteUpMorphs rather than the hand -- see 'playfield options...'. Grid origin can be specified as well as the modulus. The grid can be made visible or hidden. 2538EnvelopeEdChgs-di -- Dan Ingalls -- 13 June 2000 A couple of simple changes to the timbre editor. 2539MorphicDrawRepair-di 2540CleanBoldFix-di 2541bookAsProject-raa -- Bob Arning -- 23 August 2000 - added full-screen option to BookMorph. When in full-screen mode, there is a set of page controls floating in front of the book. - reverted BookMorph pageSize to 160@300 from 1060@800 - changed internal title in ProjectViewMorph to Helvetica and changed rectangular background to more of a drop-shadow - changed URLMorph to add a TextMorph rather than a StringMorph in response to change label... This will allow the user to make prettier bookmarks. - changed ProjectNavigatorMorph to remember user's preference for top vs. bottom edge" 2542dataIsValid-raa -- Bob Arning -- 30 August 2000 - added #dataIsValid in a couple of places where bookmorphs write themselves to remote servers since these are essentially new files, even if they are not created exactly that way." 2543pvmTweaks-raa -- Bob Arning -- 30 August 2000 - a few appearance tweaks for Alan" 2544RecolorLines-di -- Dan Ingalls -- 30 August 2000 Allows convenient recoloring of lines from the halo handle. 2545AlternateUndo-di -- Dan Ingalls -- 30 August 2000 This changeSet provides an alternate protocol and representation for undoable commands Selectors and argument arrays are used in place of blocks and parameter dictionaries. 2546UndoGroups-di -- Dan Ingalls -- 30 August 2000 Allows any morph to override its default undo behavior (needed for selections). Supports undo for border width of BorderedMorphs. Supports undo for color and border changes in multiple selections. Supports undo for move of multiple selections. Fixes a bug in display of command to undo when last command has been undone. Yet to do... [ ] undo duplicate for normal morphs [ ] undo duplicate for groups. [ ] undo delete for normal morphs [ ] undo delete for groups.

2 3

RE: A Squeak Speaklet
by Torsten.Bergmann＠phaidros.com 01 Sep '00

01 Sep '00

>funny! it works nicely (once I left the emergency evaluator and >clicked away 3 debuggers) It works here, but I didnt use Berts Plugin image. I've produced one of my own. Some tips: - starting with a clean 2.9alpha image (latest updates) - turn full screen mode off - changed a comment in a method, accepted the method, squeak will aks you for your initials. (Otherwise Squeak will ask you if code in a squeaklet is loaded/compiled and this is not intended) - file In launcher.cs and http-loader.cs from the file src.zip (go to http://ifsl.mb.uni-magdeburg.de/~michael/plugin/plugin/download.htm for download) - Open a new morphic project and enter it - Evaluate "PluginLauncher activate" - Save the image as SqueakPlugin.image into your Netscape/IE Plugin directory (you get an error while saving the image that no loader script is found, this is OK) - Note that you also need a VM, the plugin dll and appropriate plugin dll's (like squeak3d.dll if you want to use 3D) Maybe you want also prevent squeak from loading the source file (change FileDirectory>>startup for that). To test, write an sts file (startup script), like this one (tetris.sts): ---------------------------------------------------------------------------- ------- Smalltalk garbageCollect. Preferences disable: #cmdDotEnabled; compileHardCodedPref: #cmdGesturesEnabled enable: false; compileHardCodedPref: #cmdKeysInText enable: false; enable: #noviceMode. Project thumbnailFromUrl: 'http://software.freepage.de/cronos/squeak/Projects/SqueakTetris.pr' ---------------------------------------------------------------------------- ------- Then write an HTML page including the following text: <embed type="application/x-squeak-source" src="tetris.sts" width="200" height="500" align="top" memory="15" pluginspage="http://isgwww.cs.uni-magdeburg.de/~bert/squeak/plugin/download. html"> Test it in your browser. Bye Torsten

1 0

Re: Mac 2.8 Browser plugin - Help!
by John M McIntosh 01 Sep '00

01 Sep '00

on 9/1/00 8:35 AM, russell.allen(a)firebirdmedia.com at > > My currrent guess is that it had something to do with the memory being > fragmented - but since at all times the "largest available block" was > reported by the OS as being larger then 20M that seems unlikely. > Well memory might be free but the system heap might be unable to expand by the 20M that is required to satisfy the Squeak memory allocation, due to some application heap preventing the system heap from expanding. So you die even if the Browser has lots of memory, or if the finder reports a large free blocks. PS When you terminate the Squeak plugin at some point the System gives the memory back. Happily this issue all disappears with OS-X -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com =========================================================================== Custom Macintosh programming & various Smalltalk dialects PGP Key: DSS/Diff/46FC3BE6 Fingerprint=B22F 7D67 92B7 5D52 72D7 E94A EE69 2D21 46FC 3BE6 ===========================================================================

1 0

RE: A Squeak Speaklet
by Baveco, Hans 01 Sep '00

01 Sep '00

funny! it works nicely (once I left the emergency evaluator and clicked away 3 debuggers) Hans > http://wwwisg.cs.uni-magdeburg.de/~bert/squeak/plugin/speakIt/ > > -- Bert

1 0

Re: Browser plugin (was: Re: Mac 2.8 Browser plugin)
by Bob Arning 01 Sep '00

01 Sep '00

On Thu, 31 Aug 2000 16:14:50 +0200 (CEST) Bert Freudenberg <bert(a)isg.cs.uni-magdeburg.de> wrote: >Some comments, though (this applies to everyone who wants to create >Squeaklets): > >* You shouldn't use HTTPSocket to get stuff, but > (FileStream requestURLStream: 'squeak.3ds') > which asks Netscape to fetch the file. You might also look at the > PluginCodeLoader and PluginHTTPRequest stuff. Bert, Wouldn't this prevent those squeaklets from being used in Squeak that is *not* in a browser? Cheers, Bob

3 2

Re: Mac 2.8 Browser plugin - Help!
by John M McIntosh 01 Sep '00

01 Sep '00

on 9/1/00 3:05 AM, russell.allen(a)firebirdmedia.com at russell.allen(a)firebirdmedia.com wrote: > Hi, > > I have downloaded the plugin stuff to my mac, and have tried it in iCab, > Netscape and IE, but I always get a message saying: > > There is not enough memory to give Squeak the amount specified in the > 'memory' EMBED tag. > > This is on all of the test pages, all of which specify memory="15". > > I do have enough memory! I tried increasing the memory of the browser > (up to 150M for iCab) but that didn't work. I tried freeing up other > memory, but that didn't work either. I also tried changing the memory > request from 15 to other values. > > What's up? Have I missed something? > > Russell It's memory=15 not memory="15" But maybe it doesn't matter. Also if you do not code the memory attribute, Squeak attempts to allocate 20 Megabytes. The memory in question is the amount to allocate for the image outside of the working storage for the VM. This memory is allocated from the system heap, not the application heap, so increasing the memory size of your browser doesn't help. There are two messages ... dataSize = getLongFromFileswap(f, swapBytes); ... heapSize = reserveExtraCHeapBytes(desiredHeapSize, extraVMMemory); minimumMemory = dataSize + 100000; if (heapSize < minimumMemory) { plugInNotifyUser("The amount of memory specified by the 'memory' EMBED tag is not enough for the installed Squeak image file."); return null; ; } memory = (unsigned char *) sqAllocateMemory(minimumMemory, heapSize); if (memory == null) { plugInNotifyUser("There is not enough memory to give Squeak the amount specified by the 'memory' EMBED tag."); return null; ; } The first message means the amount of memory you are asking for is not large enough to load the squeak image from disk. The second message means we could not allocate the memory. Where sqAllocateMemory is: void * sqAllocateMemory(int minHeapSize, int desiredHeapSize) { /* Allocate the Squeak object heap memory from the system heap. */ #if TARGET_API_MAC_CARBON return NewPtr(desiredHeapSize); #else return NewPtrSys(desiredHeapSize); #endif } Of course NewPtrSys asks the OS to allocate the desiredHeapSize bytes from the System heap. > -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com =========================================================================== Custom Macintosh programming & various Smalltalk dialects PGP Key: DSS/Diff/46FC3BE6 Fingerprint=B22F 7D67 92B7 5D52 72D7 E94A EE69 2D21 46FC 3BE6 ===========================================================================

1 0

[BUG] Avoid text occlusion
by schwa＠cc.gatech.edu 01 Sep '00

01 Sep '00

Hi, Here is how to cause an error which sometimes crashes the system, and other times just causes the owner of the text to stop drawing and pop up a walkback. The test system is a fully updated, but otherwise vanilla, 2.9a. Embed some text in another morph (I tested with embedding in an EllipseMorph, RectangleMorph and PasteUpMorph). Tell it to 'fill its owner's shape' and 'avoid occlusions'. Then, embed another morph (tested with EllipseMorph and RectangleMorph) in the text's owner. Bring up a halo on the new morph and drag it around against its owner's top border. The error manifests itself first (that I can see) in NewParagraph>>composeLinesFrom: start to: stop delta: delta into: lineColl priorLines: priorLines atY: startingY in the following code fragment (at the '--->'): lineY > bottom ifTrue: ["Oops -- the line is really too high to fit -- back out" charIndex _ saveCharIndex. row do: [:r | lines removeLast]] ifFalse: ["It's OK -- the line still fits." maxRightX _ maxRightX max: scanner rightX. ---> 1 to: row size - 1 do: "Adjust heights across row if necess" [:i | (lines at: lines size - row size + i) lineHeight: lines last lineHeight baseline: lines last baseline]. For some reason that would not become evident to me, 'row size' is greater than 'lines size', resulting in an attempt to access a collection at index 0. Hopefully someone has better eyes than me and can figure this one out. Joshua

1 0

Jump to page: