Squeak-dev September 1998

squeak-dev@lists.squeakfoundation.org

108 participants
197 discussions

Re: String/Symbol equality
by Maloney 27 Sep '98

27 Sep '98

This issue comes up periodically. The difference results from the fact a Symbol wants equalit to mean "identical objects" whereas a String wants it to mean "collection-like objects with the same contents". To my way of thinking, the receiving object really *should* get to choose its own meaning for the "=" message, just as the it can decide what "+" or any other message means. Thus, although the asymmetry of "=" is surprising when you first run into it, I think it ends of making perfect sense when you think about it from the message passing perspective. -- John >#size = 'size' ==> false >'size' = #size ==> true > >Shouldn't they both return true? > >Symbol>>#= anObject > ^self == anObject > >Perhaps the previous method should be eliminated, and an identity shortcut >should be added to the following. > >SequenceableCollection>>#= otherCollection > "Answer whether the species of the receiver is the same as > otherCollection's species, and the receiver's size is the same as > otherCollection's size, and each of the receiver's elements equal the > corresponding element of otherCollection." > | size | > self == anObject ifTrue: [^true]. "<- IDENTITY SHORTCUT" > (size _ self size) = otherCollection size ifFalse: [^false]. > self species == otherCollection species ifFalse: [^false]. > 1 to: size do: > [:index | > (self at: index) = (otherCollection at: index) ifFalse: >[^false]]. > ^true > >--Maurice > >--------------------------------------------------------------------------- > Maurice Rabb 773.281.6003 Stono Technologies, LLC Chicago, USA

2 1

Re: v2.2 sources
by JArchibald＠aol.com 27 Sep '98

27 Sep '98

In a message dated 9/26/98 9:31:29 PM Eastern Daylight Time, dahuang(a)geocities.com writes: << Is it okay to use the 2.0 sources file with the 2.2 image? >> It's essential Jeff. Jerry.

1 0

v2.2 sources
by Jeffrey Chapman 27 Sep '98

27 Sep '98

Is it okay to use the 2.0 sources file with the 2.2 image?

2 1

Problems using #tallyMethods:
by Alejandro F. Reimondo 26 Sep '98

26 Sep '98

Folks, The simulator detects primitive failures (during simulation) when it returns #simulatorFail and halts the simulation with the title 'Code simulation error'. The problem with this technique is that we cant simulate primitive expressions returning the symbol #simulatorFail for example : ContextPart tallyMethods: [ #(1 2 #simulatorFail 4) at: 3 ] We can think that expressions like this are not used too much... But if you try to simulate code that sends the message #asSymbol to a string (begining with $s) the simbol table (array) will be traversed (sending #at:) and simulator fails! ContextPart tallyMethods: [ 'some' asSymbol ] I need to tally the methods of a (long) expression that sends #asSymbol many times, but when sending to strings starting with $s (and greater than 'simulatorFail') the simulator halts. If I open the debugger and try to restart the primitive method the SqueakVM crashes(jitter) or another walback is obtained. :-( How can I solve the problem? I have implemented a dirty hacking like.... Symbol hack: #simulatorFail with: #simulatorFait doing: [ ContextPart tallyMethods: [ 'some' asSymbol ] ] Provably a better solution exists. Ale. Content-Type: application/octet-stream; name="Symbol class-hacking.st" Attachment converted: Anon:Symbol class-hacking.st (????/----) (0000333C)

1 0

Re: Squeak Version 2.2 released
by bbennett＠unixg.ubc.ca 26 Sep '98

26 Sep '98

Whoops! The new 2.2 image continually crashes with a type 28 error ("stack has moved into application heap") on my 68k Mac (LC 475, System 7.5.5, 20MB RAM). The 2.2 beta seemed fine, but it doesn't seem to matter which VM I use, the 2.1 or new 2.2 version -- the latest image always dies with the same type 28 error, with or without extensions loaded. I can forward a MacsBug StdLog if required. -- Bruce Bennett

2 2

Re: PDA Squeak (especially Palm III)
by Maloney 26 Sep '98

26 Sep '98

At 11:37 PM +0200 9/24/98, Carsten Haerle wrote: >... >The problem with the PalmPilot is that you don't have 32bit linear address >space when you use the standard OS. Replacing it with Linux and running >Squeak on it is really an interesting technology demonstration but not >the thing most people want to do with the Pilot. > >So we need 16 bit VM for Squeak also. Having this would enable other >PDA ports as well (I don't think all PDAs have an 32bit flat memory model), >and it would make the image smaller which is interesting in the low mem >situations for PDAs. > >It should not be too difficult, as the original blue book proposed a 16bit >VM and the Squeak VM followed the blue book very closely. I would like >to see the clean interfaces to the object memory and the special code which >assumes 32bit intergers factored out, so I have the abstract interpreter and >lets say two subclasses which are named Interpreter16 and Interpreter32. We diverged farther from the Blue Book in the area of the Object Memory than in other aspects of the design. For example: a. Squeak uses direct pointers, not object table indices b. Squeak uses objects headers designed for 32-bit words c. Squeak's generational scavenger assumes a large, linear address space so it can use a simple address comparision to tell the difference between young and old objects However, I know of several object memorys designed to work well within a collection of 32K memory segments, including one designed by Mark Lentczner of Glyphic and one designed by Antero Taivalsaar (sp?) at Sun Labs, so it's certainly possible. But note that it isn't absolutely mandated by the Palm Pilot hardware architecture; the 32K memory segment limitation is entirely an artifact of their OS that I believe could be worked around with far less effort than it would take to build a new object memory (see Mike Wirth's messages to the Squeak list on this subject). If you're interested in actually writing Squeak code on the Pilot, you might be disappointed even if you had a VM than ran on it. There just aren't enough bits on that screen for programming, and the processor is not fast enough to make up for the lack of screen real estate. (If you have a fast processor, you can do things like rapid window popup/iconification to quickly shift tasks.) That's what I imagine, but I'd be delighted to be proved wrong. I love the design of the Pilot OS and would love to have one with Squeak inside it. However, the new flock of WinCE 2.0 machines have a bit more memory and processing power, and larger screens; one of them might make a lovely Squeak platform. For example, the Phillips Nino looks pretty slick. Since I don't like the WinCE applications, if I had one of these devices running Squeak I'd build my own address book and date book apps in Squeak and would seldom ever touch WinCE. -- John

2 1

Re: Toward a great Mini Squeak for PDA's
by Dan Ingalls 26 Sep '98

26 Sep '98

Folks - I have set up a directory structure on the UIUC server Small Squeaks for PDAs [ ] Mini Squeak 1.2 [ ] See_Elsewhere_Under_wince Mini Squeak 2.2 [ ] MakeYourOwn [ ] ...the files I used to make mini.image Mini.image -- 656k Mini.changes -- 1k Feel free to try out mini image. It was made from the files in the folder MakeYourOwn, and has had a fair amount of attention to reduced menus and error trapping, but I haven't really tried much other than the features I used to trim things down. It is set up to decompile the system, so you don't need any other sources. I'd be interested in hearing any comments and suggestions. Squeak is small again! - Dan

1 0

Re: Toward a great Mini Squeak for PDA's
by wirth＠almaden.ibm.com 25 Sep '98

25 Sep '98

Dan, Your plan sounds wonderful. To the extent that I'll have time in the next few weeks, I'd love to interact with you on your work. A couple comments: * As you know, the work in my group is (not yet) focused on Win-CE devices. So I'd like to exercise your builds on Palm OS devices and some experimental hardware that we have. * With respect to the porting (handwritten C) code, will your target be the Mac? That is, for each version of your Mini-Squeak, will we have the equivalent of the SqMinimal.c file that John Maloney produced several months ago? * I hope there will be some configuration options. For example, we'd like to leave the networking code in our builds. Mike Wirth

5 5

Sorry, I missent
by Randy Siler 25 Sep '98

25 Sep '98

Please pardon my last send. I replied to the list like a ding-dong. I thought Richard Harmon's little piece at the end of his last message pretty hysterical and meant to send it to a friend. My apologies Randy

1 0

Re: nil or #nil?
by Randy Siler 25 Sep '98

25 Sep '98

I just stumbled upon this at the end of a message and thought you would get a smile out of it... >Maybe I'm just getting old, pessimistic, and really grumpy. Why when I was >a kid, we didn't even have gravity yet. I had to walk 42 miles each way to >school. With a cow under each arm so they wouldn't float away. And I >walked home for lunch. Which was dirt, but only when we got rich enough to >afford dirt! Which was invented when I was in the third grade. And third >grade was the best 6 years of my life . . . .

1 0

← Newer
1
2
3
4
5
6
7
8
...
20
Older →

Jump to page: