Squeak-dev April 2000

squeak-dev@lists.squeakfoundation.org

141 participants
506 discussions

(Very) Newbie Questions
by Mark 05 Apr '00

05 Apr '00

Hi! A friend of mine pointed out Squeak to me and so I've downloaded it and have played with it for a bit. It looks very cool and, thanks to some of the tutorials available, I've been able to construct some classes and play with them. I seem to have run into a mental block while playing with Squeak. Are there any good, basic introductions to Squeak/SmallTalk that would explain some of the fundamentals (variables, keywords, etc) that would help me on my way? Also, the philosophy seems to be much different from more "traditional" languages. For example, when I close all the "PlayWithMe" windows, how do I get them back (without going back to the original image)? How do I determine which classes create a complete program that can be "launched"? (Also, is the correct way to "launch" things via the Workspace?) I guess, in short, are them some guidelines that will help me learn this stuff quicker or is it all by trial and error? Thanks for taking the time to read the ramblings of a Squeak newbie... -- Mark -- Mark marquedej(a)onebox.com - email (760) 708-6400 x1067 - voicemail/fax __________________________________________________ FREE voicemail, email, and fax...all in one place. Sign Up Now! http://www.onebox.com

2 1

Re: [Q] executing SOME compiledMethod (classless)
by Lex Spoon 05 Apr '00

05 Apr '00

How about using blocks instead of CompiledMethod's? Blocks can be evaluated by sending #value. You could make the first argument of the block be the "receiver". -Lex "Peter Novak" <pno(a)whitestein.com> wrote: > :>>> Bob ;>>> > > Workspaces were just an example I used for explaining our problem. I should > say that I want to create something like better morphs. We are working on > prototypes in Squeak. And prototypes do not have ANY class. Morphs are also > prototypes, but we don't like the idea of having separate class for each > morph only because it has some scripts. What about having objects like > morphs with their scripts connected between each other and really working? > And now you come to the same problem, because you need to execute scripts > for these prototypes (even we need it concurrently). Therefore we were > looking for some clear solution, like using some primitive which will > execute such a script. But it seems that we have to use some dirty trick > (like that with class like UndefinedObject). The question is only to use the > most clean from those dirty tricks ;>. John Maloney (I hope he's the one who > implemented Morphs, I am sorry if I am wrong) solved this problem in Morphs > with creating class for each morph. We are looking for some better solution. > We want to have in our prototype based system as few bindings to Smalltalk > class system as it will be possible. That's explanation of our problem. > > We used workspaces for explaining our problem, because we wanted to explain > the problem as simply as it was possible. It seems now, that we made not > very good example of that what we want. Nevermind. It's okay now, I think. > > Regards, > > Peter. > > > > -----Original Message----- > From: Bob Arning <arning(a)charm.net> > To: squeak(a)cs.uiuc.edu <squeak(a)cs.uiuc.edu> > Date: 31. marec 2000 16:52 > Subject: Re: [Q] executing SOME compiledMethod (classless) > > > >On Fri, 31 Mar 2000 16:22:42 +0200 "Peter Novak" <pno(a)whitestein.com> > wrote: > >>But imagine concurrent Workspaces where SOMEONE (doesn't matter now who - > >>maybe people working in distributed environment) want to evaluate his own > >>expression in his own workspace. Now You have a problem, because you are > not > >>able to say whose #DoIt will be executed at the moment. > > > >Peter, > > > >I suspect that this is far from the only problem you will encounter with > multiple users sharing one image like that. Too much stuff is not thread > safe. One option for the doIt problem (and you did touch on it) would be for > each user to have a private class in which context doIts are evaluated (i.e. > evaluate the code in PeterNovaksPrivateClass or BobArningsPrivateClass > rather than in UndefinedObject). But, as I said, this may be the least of > your problems. > > > >Cheers, > >Bob

2 1

Re: VNC Server
by Marcel Weiher 05 Apr '00

05 Apr '00

> 2. While most Canvas commands are easy to implement, > #paragraph:bounds:color: is a notable exception. I notice that the > Postscript guys had to implement some sizable code for this. :) Sure did! :-) It was also left in its <euphemism_alert> somewhat suboptimal </euphemism_alert> state because the morphic drawing subsystem was to be revamped by Squeak Central at the time. Some of those changes seem to have happened in the meantime, but the full overhaul was apparently never completed. > I extended their work into a more general "CanvasCharacterScanner" class. Great! Maybe we can start work on getting a modern text-display subsystem working. It looks like most of the elements are now in place. > Instances of this class can be used to transfarm a "draw paragraph" > command into a sequence of simpler "draw text" commands. This single > bit of code ought to be able to replace all other implementations of the > "draw paragraph" command, In fact, I've been using it as a replacement > for FormCanvas's default implementation for a while, and it works just > fine. Surely it would work for Postscript as well. I'll see if I can throw away my hacks, and maybe improve the fidelity of text output somewhat. Does this also work for text-on-curve? Marcel

1 0

RE: (Very) Newbie Questions
by Torsten.Bergmann＠phaidros.com 05 Apr '00

05 Apr '00

See http://www.phaidros.com/DIGITALIS/englisch/lit/lit00002.htm for my literature list on Smalltalk books. You can also read the first chapters of a book about Squeak at http://coweb.cc.gatech.edu/squeakbook/3 The upcoming Squeak 2.8 release will have support for image segments (projects you can export/import into your image). Maybe we can create some projects teling people how to use Squeak and Smalltalk. Torsten -----Original Message----- From: Joshua Gargus [mailto:gargus@cs.ualberta.ca] Sent: Mittwoch, 5. April 2000 08:31 To: squeak(a)cs.uiuc.edu Subject: Re: (Very) Newbie Questions Hi Mark, Welcome to Squeak! For the nuts and bolts of the language itself, I'd recommend "On To Smalltalk" by Patrick Henry Winston. It gives a good introduction to the syntax of the language as well as common idioms. Don't worry about the graphical examples later in the book, as Squeak's display paradigm (Morphic) differs significantly, and is much more fun to use! Morphic (and other) documentation can be found on the main swiki at http://minnow.cc.gatech.edu/squeak.1 This is a good place to go to see what's going on in the Squeak community. Have fun, Joshua

1 0

Re: (Very) Newbie Questions
by Steve Elkins 05 Apr '00

05 Apr '00

"Mark" <marquedej(a)onebox.com> wrote: > I seem to have run into a mental block while playing with Squeak. Are > there any good, basic introductions to Squeak/SmallTalk that would explain > some of the fundamentals (variables, keywords, etc) that would help me Please have a look at Andrew Greenberg's quick reference http://www.gate.net/~werdna/squeak-qref.html and see if it meets (at least some of) your needs. Welcome to Squeak, Steve

1 0

[ANN] ANSI Messages & Tests Updated (long)
by R. A. Harmon 05 Apr '00

05 Apr '00

I have updated my ANSI Messages and their SUnit tests to add, and change non-conforming, all (eventually) messages required by the ANSI Smalltalk Standard. Each file is installed or filed into its own change set so almost all changes to Squeak can be determined fairly easily. I appologize for my idiosyncratic code formatting. ANSI Standard Messages: Finished Numeric Protocol messages. Finished Date and Time Protocol messages. Added RuleDate, RuleIndexDate, and RuleSelectionCodeDate support classes. The RuleDate and sub-classes instances are suitable for representing dates such as the start of daylight saving time in the USA (last Sunday in November); Thanksgiving Day (fourth Tuesday in November) in the USA; and the last day of daylight saving time (the Saturday before the last Sunday in October) in the USA. They have a quick and dirty implementation which probably could be done better. There is slightly different behavior of some ANSI methods compared to the current Squeak versions. It can cause some subtle errors like that I corrected in RunArray class>>#scanFrom: where it scanned for the fontNumber using Number>>#readFrom: when it was looking for an integer (I'm pretty sure). The syntax of an ANSI number is different and confused the scan, so needed to use Integer>>#readFrom:. I never did find a way to determine the value for DateAndTime class>>#clockPrecision so I bailed out and returned one second. I also didn't get ANSI Fraction behavior exactly correct with respect to reducing the fraction. I only seems to affect a couple of messages and a fix I did not include seemed to involve quite subtle interaction which sends the system into an infinite loop of coercion. I got tired of futzing with it and the time I was spending seemed to be disproportionate to the problem. It can be sorted out later -- sorry. ANSI Smalltalk Standard Type Protocols: Added ClassDescription >> #definedAsProtocolNames to determine what instances directly conform and use in testing. SUnit with Protocol Assertions: Finished implementing TestCaseANSI >> #value:should:conformTo:selector:opRECEIVER:. ANSI Standard Messages Unit Tests: Commented out ObjectClassANSITest >> #testXallSuperclasses as I could not tell what class must be present in ANSI. Commented out ExceptionANSITest >># testXresignalAsX as it hangs SUnit tests by resig. ex., ex. handle by resig. ex.--dumb. Finished all but super protocol Numeric Protocol tests. Finished all but super protocol Date and Time Protocol tests. The "all but super protocol ... tests" refer to messages like Object>>#copy. I have not added these mostly Object message tests. My tests of the ANSI standard messages are sort of lame but seem to me a good start, and are were of great use to me in catching errors and the subtle interaction of the slightly different behavior of some ANSI methods compared to the current Squeak versions. My tests of some of the Exception messages are really poor, and a couple have no test just a commented out start of one. They are now available at my web page: http://homepage2.rconnect.com/raharmon/ -- Richard A. Harmon "The only good zombie is a dead zombie" harmonra(a)webname.com E. G. McCarthy Spencer, Iowa

2 2

Re: [ENH] Flexible LowSpaceWatcher
by Lex Spoon 04 Apr '00

04 Apr '00

This is tres neato. A nice, simple improvement, IMHO, would be to have more than one level of severity: 1. Discard completely unneeded things. 2. Discard things that are unneeded, but expensive to recompute. 3. Discard things that are irreplacable. Unfortunate, but probably better than letting the image just quit. "Space hogs" would register with one of these levels, and the low-space code would try less-severe hogs before moving to more severe ones. Food for headaches: how do you extend this kind of scheme to handle other resources, such as file handles and sockets? -Lex Georg Gollmann <georg.gollmann(a)tuwien.ac.at> wrote: > --============_-1257709381==_============ > Content-Type: text/plain; charset="us-ascii" ; format="flowed" > > The current LowSpaceWatcher just alerts the user. This is not useful > for servers or for situations when objects are cached and should be > discarded when space becomes tight. I have now generalized the > mechanism used in my MailArchiver > (http://macos.tuwien.ac.at/Squeak/mailArchiver.html).

1 0

Re: VM Query: Contexts on Stack?
by Allen Wirfs-Brock 04 Apr '00

04 Apr '00

At 09:54 AM 4/4/00 +0000, Lex Spoon wrote: >... >Hmm. There still seem to be a lot of cases where the compiler *can't* >catch it. For a common example, how would a Smalltalk compiler decide >about this one: > > someUnknownObject do: [ :x | ... ] > > >The block is almost definately not captured, because methods called do: >rarely capture their block. But how can a compiler figure this out? It >seems that you would still want to have the dynamic flush-to-heap >methods described earlier, so that you can optimistically stack-allocate >in cases like this one and then back out of it if the block turns out to >get captured. The issue is not whether the block is captured but whether the block references any variables from enclosing scopes. Any such variables will have lifetime exceeds that of the block activations and the associated method activation and hence must be heap allocated.. In the above example, the only variable that is shown is a block argument. Such a variable would normally have a lifetime that is the same as the block activation (assuming that it is not referenced by any retained nested blocks) so it can be stack allocated in the activation record of the block. If the example had been: someMethod |captured someUnknownObject| someUnknownObject do: [:x| captured someMessage] Then the method temp named captured could have a lifetime that extends beyond the method activation. This variable needs to be heap allocated. Note, however, that the variable someUnknownObject isn't "captured" by the block so it can be stack allocated in the method's activation record. Allen_Wirfs-Brock(a)instantiations.com

1 0

Multiple native windows? RE: VNC Server
by cgundel＠domainpharma.com 04 Apr '00

04 Apr '00

>On the other hand, there are several useful things that could be done in image >code if you want to avoid that. Certainly a plugin for the compression seems >sensible. Assuming you want to pass a commandline parameter in to specify VNC >instead of local display (you could also do it quite nicely just by using a >config file via the Smalltalk>readDocumentFile method) you could create an >instance of VNCDisplayScreen instead of a DisplayScreen. Override #beDisplay so >you don't call the prim (the VM can easily be modified so that no window is >even created until & unless the prim is called - it works very nicely on Acorn) >and add something like #doVNCDisplay to the class and a null implementation to How hard would it be to take this mechanism a step further so that multiple OS windows could be managed by a Squeak image? This would go a long way towards providing VisualWorks style window features. I'm practically drooling. ;-) -Carl

1 0

Re: T-gen in Squeak?
by Dan Ingalls 04 Apr '00

04 Apr '00

>Russell Allen wrote: >> >> Hi all, >> >> Has anyone ported the T-gen parser generator to Squeak? NOTE: STP just responded (with a 120K attachment), but the message was dated 1904, so you might not see it unless you scroll for it. - D

2 1

Jump to page: