Squeak-dev May 1998

squeak-dev@lists.squeakfoundation.org

82 participants
182 discussions

Re: First automatic updates to beta image available
by Sabine van Loon 20 May '98

20 May '98

Bijan wrote: > > It seemed reasonably fast to me over my 33k connection. How big were the files? > Sorry guys, I was working on my computer which has a 14k4 modem. I normally e-mail on Sabines machine (that's why you get 'from Sabine van Loon') which has a 28k8 modem. In less than a week the cables for an ISDN line will be there and then I will flame through the Internet ;-) Well, sort of. Take care, Reinier.

3 3

Re: squeak 2.0
by Maloney 20 May '98

20 May '98

>Sorry about this. The message below was supposed to go to the sender, >not the mailing list. I really need to remember to check those >headers before sending. > >I wrote: >> Hi, >> >> It looks like this email got sent to me directly somehow or >[deletia] > > >Oh well, since I've wasted bandwidth alreadyy, I may as well try to >tack on useful information as well. > >I tried running the 1.31 VM on the 2.0 image and got another segfault, >so it appears to me that either the image got mangled ad download time >or my binhex converter doesn't work. > >Binhex and StuffIt aren't too common outside of the Macintosh world. >It would be nice if somebody with access to the required software >could repackage the distribution as a tar.gz or even a .zip file. I think you can find all the source files uncompressed at: ftp.create.ucsb.edu/pub/Smalltalk/Squeak/2.0Beta/VM Thanks to Stephen Pope for making them available. -- John

1 0

How new users get Squeak
by Ted K. 20 May '98

20 May '98

Folks, When a new user of Squeak comes to the homepage (squeak(a)cs.uiuc.edu), he or she just wants to download a working system. The instructions in the Download section are divided by machine and OS, but do not clearly state how to download the correct files. As organizer of the pages, I would like some volunteers to help with the instructions for each OS. Ideally you are a relatively new user, who has been through the process of getting the files recently. You are very familiar your OS, and how to get Squeak started on it. You have an eye for what might go wrong and what might be misunderstood. Please volunteer to edit the section on downloading Squeak for your OS. Supply the correct pointers to the files the user will download from UIUC. Instructions for the mechanics of downloading (binary? zipped?) are vital. There is a general Trouble section, but we may need specific trouble shooting instructions for each OS. Reply to <tedk(a)wdi.disney.com> with what OS you would like to write for. I would love to have the instructions be foolproof and crystal clear. Thanks! --Ted. Ted Kaehler, Walt Disney Imagineering, R&D (home) 3415 Cork Oak Way, Palo Alto, CA 94303. voice (650) 424-1070 http://www.webPage.com/~kaehler2/ Give me your old Macintosh! I will put it in a first grade classroom.

1 0

Re: Re: MessageTally
by sqrmax＠cvtci.com.ar 20 May '98

20 May '98

Hi. >>Ahhhh... but it has happened again and yes! (well... no!), startTime is nil >>when printed in the debugger. >Another possibility is that the MessageTally is still running while >you are looking at it's stack with the debugger. Nono... this is not the case. I check the error messages after the tally opens the report. >Or perhaps >the reverse; it is NOT running but the startTime variable only >has a meaningful value when it *is* running. Hmmm... but then why the error box? >We use MessageTally quite a lot, so I think it generally works. >What was it that you were trying to do? I run a MessageTally on aBlock, and from time to time, after the tally opens the report, I discover a covered error box about nil not understanding #-... it looks as if startTime is nilled somehow. Andres.

1 0

RE: couple o' questions
by David Farber 20 May '98

20 May '98

i changed double to float and still was getting errors. i am still not convinced that it is handling the underscore (or api names in a string) correctly. dave At 04:20 PM 5/19/98 -0000, you wrote: >David, >Float arguments are not implemented, yet. >(See ExternalInterface>>#adaptArgumentDouble under 'arguments' category) >:-( >Provably in the near future I will have some time to implement it. >Try to use long integers to code voltages. > e.g.: (realVoltage * 100) truncated >I know it is not fun, but it will work. >Ale. > >---------- >Desde: David Farber[SMTP:dfarber@numenor.com] >Enviado el: Martes 19 de Mayo de 1998 09:36 >Para: squeak(a)cs.uiuc.edu >Asunto: RE: couple o' questions > > >>API call names can be symbols or strings. Try this >> >>aoVWriteDevice: deviceNumber channel: channelNumber voltage: voltage >> <api: 'AO_VWrite' ushort ushort double short> >> ^self invalidArgument > >doesn't work. i get an "Invalid Argument" error. when i inspect the call >stack, i see that <deviceNumber> is set to self, <channelNumber> is set to >what <deviceNumber> should be, and <voltage> is set to what <channelNumber> >should be. > >dave > >-- > j. david farber > oo architect+mentor >numenor labs incorporated >in sunny boulder colorado > dfarber(a)numenor.com > www.numenor.com > > >Attachment Converted: c:\eudora\dfarber_num\attach\RE couple o' questions2 > -- j. david farber oo architect+mentor numenor labs incorporated in sunny boulder colorado dfarber(a)numenor.com www.numenor.com

1 0

Re: KeyUp events
by jecel＠lsi.usp.br 20 May '98

20 May '98

"Ted K." <tedk(a)wdi.disney.com> wrote: > > [event graph idea] > I too have been feeling the need for something > like this. Any ideas on how to keep from computing > the whole graph for each event? This is only a problem for the "push" arcs of the graph, not the "pull" ones. My solution would have a filter (pushInput/pushOutput kind of component) disconnect from its input (unsubscribe) whenever it had nobody linked to its output and reconnect (subscribe) when a new customer arrived. I think that this (and what you describe below) would avoid having events flow though any more of the graph than strictly necessary. Buffers are a problem (push/pull kind of objects) because they stay subscribed and grabbing events all the time, but I don't think anything can be done except not putting them where they are not really needed. > I picture the panes of a browser attaching and detatching > themselves from the keyboard input pipe as > you move the mouse over them. That is the idea. It might be best not to take this too far, however. Windows could work like this and yet have a single "demux" object distrinute the events among the subpanes (or submorphs). This would be a pull/push object - what I have called a "pump". This scheme has the advantage of being compatible with a lot of existing code. > --Ted. -- Jecel

1 0

Numeric problem?
by gr＠fuzzy.uba.ar 20 May '98

20 May '98

Does anybody has this problem: try doing -1e10/1 ==> error: LargeNegativeInteger doesNotUnderstand highBit. Actually this is not the message, but it's something equivalent (I can't check it now or here) To fix it I needed to write: LargeNegativeIntegera>>highBit ^ self negated highBit and change something in SmallInteger>>highBit. I'm not sure if this is a good solution, but it seems to work for me. BTW: Why is there a self isNegative ifTrue: [] in LargePositiveInteger>>highBit (shouldn't it be in LargeNegativeInteger?, whithout the self isNegative..) Buggy Bye! Richie++

2 1

SqueakMarks and G3 Powerbooks
by Maloney 19 May '98

19 May '98

For some time, I've been in the habit of carrying around a floppy with a stripped down Squeak and VM (not compressed!) that I can pop into any Mac with a floppy drive to measure its performance on a couple of simple, non-graphical benchmarks. Below are are the results for a number of machines. The new G3 series PowerBook at 250 MHz is close to the top in performance. Send speed in sends/sec: | r t | t _ Time millisecondsToRun: [r _ 26 benchFib]. r // t * 1000 636000 637000 636000 -- G3/266 (tower) 636000 636000 654000 -- 8600/300 604000 604000 604000 -- PowerBook G3 250 MHz 561000 561000 561000 -- 6500/300 561000 561000 547000 -- G3 (desktop) 535000 535000 548000 -- uMax 600x (603e/280, 1 MByte inline L2) 481000 491000 462000 -- 20th Anniversary Mac 436000 428000 420000 -- 6500/225 351000 346000 346000 -- 3400 341000 332000 341000 -- 2400c 327000 318000 322000 -- 1400c/166 196000 199000 199000 -- John's PowerBook 1400c/117 178000 178000 177000 -- John's Mac 8100/100 Bytecode speed speed in bytecodes/sec: 10000000 // (Time millisecondsToRun: [20 benchmark]) * 1000 14992000 15015000 15384000 -- G3/266 (tower) 14285000 14992000 13947000 -- 8600/300 14285000 14285000 14285000 -- PowerBook G3 250 MHz 13333000 13623000 13333000 -- G3 (desktop) 12500000 13037000 12771000 -- 6500/300 11111000 12500000 11764000 -- uMax 600x (603e/280, 1 MByte inline L2) 10718000 10718000 10718000 -- 20th Anniversary Mac 9372000 9233000 9680000 -- 6500/225 8103000 8116000 7898000 -- 3400 7898000 7886000 7898000 -- 2400c 7142000 7057000 6451000 -- 1400c/166 4725000 4761000 4723000 -- John's PowerBook 1400c/117 4166000 4166000 4110000 -- John's Mac 8100/100 Note: Each test was run three times; inconsistency across a set of runs means that either some other application was stealing time from Squeak or that some odd caching phenomena makes performance inconsistent, even on an otherwise idle machine (I've seen this most often on PPC603-based Macs without an L2 cache.) The G3 is a model of consistency! -- John

1 0

11 new updates to the beta image
by Dan Ingalls 19 May '98

19 May '98

F.Y.I. Herewith a summary of bugs/features addressed. Please send me a *private* reply if you are tracking updates. I will stop broadcasting these messages if only a handful of people are doing this. Thanks - Dan I ------------------------------- Change Sorters now work in Morphic Auto-select fixed in messageLists (ie browse senders) Broadcast to multiple update servers reinstated (we will reinstate for update clients after test) Debugger pc highlight updates when method is same Transcript now deletes from top not bottom (duh) Copy/paste in Morphic Fix for world halo creep Display flash: fix from Andreas Raab Float comment fix from Tim Olson Morphic Windows now obey ActiveOnlyOnTop, as in MVC Cleaned up chooseColor; eliminated the ColorChart Misc. small fixes from Georg Gollmann, Tim Olson, Andreas Raab Can now paste book pages (any pasteup/playfield/holder) into books from the book control panel Numerous fixes from Scott W, including... * Removed the last residual linkage between MorphicModel and anything in the "PasteUpMorph and Player" regime. Recently, the MorphicModel had only been used for the purpose of enumerating all the extant Players. Now this is done via a cache held on to by the World and recomputed when necessary by scanning the Morph tree. * Combined the "embed" and "place in..." commands into a single command, called "embed", similar to the original "embed" but able to work in the intuitively obvious way even when the operands are nested any number of levels deep in PasteUpMorphs. * Removed the ambiguity about where an object's name resides, and consequently removed the #externalName slot from the ActorState. Now the external name lives *only* as the value of the #name property. It gets automatically transported back and forth between the flexmorph and the real morph in the same manner as the ActorState. * "duplicate" and "make another instance of same user class" are now on much sounder footings. You can script an object and either duplicate or new-instance it, and may find that the newly-minted copy works as you'd hope. In the case of multiple instances of a single user class, it's even robust about post-editing of the class, which is to say: you can add instance variables, and change their type, at will, and all extant instances should "catch up" intelligently. * Merged my recent menu changes with John's recent "copy/paste morph" addition. And, because of the above, "copy/paste morph" now works beautifully even with *scripted* objects. Among the bugs fixed and glitches addressed: * cmd-t no longers gets ParagraphEditors into an intractable state. * Viewer in EToy no longer vanishes after you submit a new name. * Several long-standing errors relating to Bookmorph menus fixed. * Spruce-ups of appearance and default properties of BookMorphs in Parts Bin. * BookPages start animating as soon as you go to them now. * Newly added Bookpages now start out with size, border, color as predecessor. * The paint-invoking-widget in EToys now works the same as in general Morphic. * No longer drop into debugger when trying to get to Studio Cover (what is that, anyway?). * Fixed several "name" bugs relating to flexing. * When a PasteUp becomes a PartsBin, its event handlers are suspended. * Empty slot & method banks are allowed; allows much smaller Viewers. * Long delays when duplicating e.g. StopButton are no more. * Holder's cursor no longer freezes at one. * Bug fix in & fine-tuning of available-costumes in the change-costume menu. * The old "userStuffBook" is reconstituted to use PasteUpMorphs for pages. (note: it had ClippingMorphs in it -- a class never referred to in the image, but surviving in ancient instances held on to by a class variable!) Some known residual problems: * Newly-ticking scripts sometimes are unresponsive to Stop button. (workaround: hit Go, *then* hit Stop). * Mouse-over halos not appearing over objects on Playfield in etoys. * Seem to have lost content of Imagis tab in etoy framework. * Names sometimes too eagerly assigned to innocent objects.

2 2

RE: couple o' questions
by Alejandro F. Reimondo 19 May '98

19 May '98

David, Float arguments are not implemented, yet. (See ExternalInterface>>#adaptArgumentDouble under 'arguments' category) :-( Provably in the near future I will have some time to implement it. Try to use long integers to code voltages. e.g.: (realVoltage * 100) truncated I know it is not fun, but it will work. Ale. ---------- Desde: David Farber[SMTP:dfarber@numenor.com] Enviado el: Martes 19 de Mayo de 1998 09:36 Para: squeak(a)cs.uiuc.edu Asunto: RE: couple o' questions >API call names can be symbols or strings. Try this > >aoVWriteDevice: deviceNumber channel: channelNumber voltage: voltage > <api: 'AO_VWrite' ushort ushort double short> > ^self invalidArgument doesn't work. i get an "Invalid Argument" error. when i inspect the call stack, i see that <deviceNumber> is set to self, <channelNumber> is set to what <deviceNumber> should be, and <voltage> is set to what <channelNumber> should be. dave -- j. david farber oo architect+mentor numenor labs incorporated in sunny boulder colorado dfarber(a)numenor.com www.numenor.com Content-Type: application/ms-tnef Attachment converted: Cog:Untitled 3 (????/----) (0000D10A)

1 0

← Newer
1
...
7
8
9
10
11
12
13
...
19
Older →

Jump to page: