Squeak-dev April 1998

squeak-dev@lists.squeakfoundation.org

107 participants
279 discussions

Re: compiling blocks
by David Farber 29 Apr '98

29 Apr '98

>> how do i compile '[ :foo | foo bar]' on the fly so that i can store it in >> an instance variable and send value: to it sometime later on? > >Have a method returning this block. E.g. >fooBlock > ^[ :foo | foo bar ] > >otherMethod > object setBlock: self fooBlock no, i have to create the string '[ :foo | foo bar]' on the fly. (because i don't know bar). i need to create this block at *run time*. dave -- j. david farber oo architect+mentor numenor labs incorporated in sunny boulder colorado dfarber(a)numenor.com www.numenor.com

2 1

compiling blocks
by David Farber 29 Apr '98

29 Apr '98

how do i compile '[ :foo | foo bar]' on the fly so that i can store it in an instance variable and send value: to it sometime later on? also, what if i reference self in the block, like '[ :foo | foo bar: self]'. what do i need to do to make sure that self evaluates to the object that is holding the block? thanks, dave -- j. david farber oo architect+mentor numenor labs incorporated in sunny boulder colorado dfarber(a)numenor.com www.numenor.com

2 1

Re: compiling blocks
by David Farber 29 Apr '98

29 Apr '98

>The "block := ..." statement is the code... I just sent the #value: message >at the end for an example. You could have returned the block instead of >evaluating it. For example, >Integer>>foo > | test | > test := 5. > ^Compiler new > evaluate: '[:a | a + test + self]' > in: thisContext > to: self > notifying: nil > ifFail: [nil] >will return a block content that you can then send #value: to (4 foo value: >10 ==> 19)... ah, yes. i was throw off by "evaluate". >If you don't want access to variables declared outside the block, you can >simply use: > Compiler evaluate: '[ :foo | foo bar]' actually, that is precisely what i wanted! thanks!! dave -- j. david farber oo architect+mentor numenor labs incorporated in sunny boulder colorado dfarber(a)numenor.com www.numenor.com

1 0

Re: compiling blocks
by David Farber 29 Apr '98

29 Apr '98

>Well, you could try > > [:foo :instruction | foo perform: instruction], > >then call the block like > > thatBlock value: someObject value: #bar. > >Or, if you need the possibility of multiple arguments: > > [:foo :instruction :arguments | foo perform: instruction > withArguments: arguments] > >And call it like: > > theOtherBlock value: someObject > value: #inject:into: > value: (Array with: 1 with: enumBlock). > > >These are a little simpler than using Compiler, but can allow a fair >amount of generality. hmmmmm....what kind of performance are we talking about here? just looking at the simple cases where i want to do this, it looks like an easy half-dozen invocations of the block. in the more common, complex cases, it looks like there could easily be hundreds or thousands of invocations. that's why i'd like to do the compiling up front. dave -- j. david farber oo architect+mentor numenor labs incorporated in sunny boulder colorado dfarber(a)numenor.com www.numenor.com

1 0

Blue button under Unix
by Sabine van Loon 29 Apr '98

29 Apr '98

Folks, when using Squeak under AIX via Exceed (a Windows95 X terminal client) the blue button is available by using Ctrl+Left mouse button (red button). The Windows95 implementation uses right mouse button for blue button. Does anyone know if this is the normal behavior under Unix? Or could it be the X windows client? Or does AIX do things differently in that area? Reinier.

1 0

Linux squeak
by Alexander Lazarevic 28 Apr '98

28 Apr '98

Hi! I just joined the mailinglist a few days ago. There was a mail from someone (sorry I accidentally deletetd it) complaining about the lack of Linux squeak to produce sound. I would like to take the opportunity to point out some other shortcomings of Linux squeak. The one thing, that really annoys me is that squeak won't interrupt on CTRL-C. If I get stuck in an endless loop I have to kill the squeak process, restart squeak and replay the changes. Mentioning starting squeak reminds me that I have tried to start squeak with the -xshm option to enable X shared memory extensions. If I try to resize the squeak work window it crashes. This won't happen when I start squeak without -xshm. What confuses me is that some of the sourcecode in the image uses CR as the lineend character. Everytime I want to have a look at a method which has this line terminator I just get garbage in the method view. I have to do a format and accept to get it straight. Shouldn't the display method treat LF and CR equally? Other thing I noticed, is that output to Transcript is very slow. Something like <? 80 timesRepeat: [Transcript show: '#'] ?> really gives you a coffee-break. I don't know if this is just a linux specific problem. After all that bashing I want to make clear, that squeak has some aspects I really like. Just to pick one: PWS is a very ... "cool" thing. :) I appreciate the rapid development of squeak and that it offers a playground for new ideas. This made me return to it. (After first contact with squeak about one year ago, my first reaction was to set up "The official Squeak haters page" at our university. -- Things can change :) Alex. -- ------ Alexander Lazarevic ------------------------------------------------ -- Otto-von-Guericke-Universitaet mailto: lazarevi(a)cs.uni-magdeburg.de -- -- Computational Visualistics http://www.cs.uni-magdeburg.de/~lazarevi --

4 5

Re: Smalltalk <-> Objective-C
by Tim Rowledge 28 Apr '98

28 Apr '98

On Tue 28 Apr, David Stes wrote: > For example, with the question on QuickTime-support that we had a while > ago on this list, of making a "link" between Squeak and QuickTime, it was > clear that the Squeak people feel, that having such external modules, is a > bit against their philosophy of being able to access everything, from top > to bottom, as Smalltalk code. (well, there was also a concern of > portability, but that concern wouldn't apply in the ST<->OBJC case). I'm not sure how you got this impression David, but it certainly isn't one I got; several of us that look after the assorted ports have created prototypes of calls from Sq to outside modules precisely because it seems such an obvious idea .. I _think_ that what Dan wrote expressed an opinion that there is no point in adding special QuickTime primitives to Sq since it is not portable and not a core requirement for the system. I believe he further suggested exactly the genera lized external calling capability that I refer to above. tim -- He who hesitates is probably right. Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 856-7230 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

1 0

Re: Smalltalk <-> Objective-C
by Carl E Gundel 28 Apr '98

28 Apr '98

How about the idea that instead of translating the Squeak VM from Smalltalk to C, that it would be translated to Objective-C? Wouldn't it then be easier to write the VM in a a more natural style of Smalltalk? Is Objective-C standard enough to make this a feasible idea? -Carl On Tue, 28 Apr 1998, David Stes wrote: > On Tue, 28 Apr 1998, Steve Dekorte wrote: > > > Is anyone working on a Smalltalk to Objective-C interface? > > I don't mean code conversion, I mean a direct interface between > > running Smalltalk and Objective-C code. > > My first impression was that it might be a good idea, but then later I was > thinking that it is against the idea, both of Squeak and Objective-C. > > For example, with the question on QuickTime-support that we had a while > ago on this list, of making a "link" between Squeak and QuickTime, it was > clear that the Squeak people feel, that having such external modules, is a > bit against their philosophy of being able to access everything, from top > to bottom, as Smalltalk code. (well, there was also a concern of > portability, but that concern wouldn't apply in the ST<->OBJC case). > > Anyhow, I certainly respect this feeling, and when you see how nice Squeak > is, and how much they achieved, like how easy you can get Smalltalk from a > different source, to work in Squeak (it would be difficult if you have > external modules, I think). > > Objective-C is, in its heart, runtime support for a Smalltalk -> C > translator. It's better to just continue to do it like that, I think, > (i.e. conversion) and make sure that we follow the standard that Squeak is > setting. > > David. > ------------------------------------------------------------------ Carl Gundel carlg(a)world.std.com Shoptalk Systems 508-872-5315 author of Liberty BASIC, a 1996 PC Magazine Awards Finalist! http://world.std.com/~carlg/basic.html ------------------------------------------------------------------

2 1

Not too slow?
by Sunghun Seo (Sonny) 28 Apr '98

28 Apr '98

For few weeks, I looked around Squeak very serioulsy. And I am very satisfed with its basic theory. But, when I face the overall performance, especially speed, it is hard to say that it is worth enough. Therefore, I am trying to find the reasons. Maybe, have Squeak inherited from the notoriously slow Smalltalk 80? or is it already the best condition to perform on various platform? Could you please give me some advice that what makes Squeak so slow? PS. I am running Squeak on Intel platform. Sonny Seo sseo(a)otto.bf.rmit.edu.au

3 2

Re: [Q] Ask on String manipulation
by David Stes 28 Apr '98

28 Apr '98

glenn wrote: > > As you suggest, this is used a lot in error handlers and other places, > especially in the face of internationalization. > > I don't think it's a huge project to create this functionality in Squeak, > and while you can't steal the VW code, I think we'd benefit from having > compatible syntax, so I'd encourage you or some other Squeaker to check out > the VW description. It's usually used like in [self error:"No object at offset %i !",offset] in Objective C. Talking about error handling, I've renamed our -ifException: to be like Squeak's -ifError: to be compatible.

3 2

Jump to page: