Squeak-dev August 1998

squeak-dev@lists.squeakfoundation.org

99 participants
188 discussions

Start a nNew thread

Re: nil or #nil?
by JArchibald＠aol.com 25 Aug '98

25 Aug '98

BTW, there shouldn't be any questions about self, should there? I mean, self refers to a variable. Jerry.

2 1

Re: nil or #nil? (What the ANSI Standard says)
by JArchibald＠aol.com 25 Aug '98

25 Aug '98

In a message dated 8/25/98 2:55:07 PM Eastern Daylight Time, Allen_Wirfs- Brock(a)Instantiations.com writes: > If an identifier appears > as an <array element> and it is one of the reserved identifiers nil, true > or false the value of the corresponding element of the collection is the > value of that reserved identifier. so it looks like the standard says that #(nil true false) is equivalent to Array with: nil with: true with: false Is that the way others read it? Jerry.

1 0

Re: nil or #nil?
by JArchibald＠aol.com 25 Aug '98

25 Aug '98

Doesn't VisualWorks require the combined form, i.e. #(#a #b #c) If so, what does it do with #(nil true false)? Jerry.

1 0

Re: nil or #nil?
by JArchibald＠aol.com 25 Aug '98

25 Aug '98

The problem is not that the things referred to by nil, true and false can't be viewed as literals. The problem is that #nil, #true, and #false are symbols, and array defs would have #(nil true false) be equivalent to Array with: 'nil' asSymbol with: 'true' asSymbol with: 'false' asSymbol without complicated things in the grammar. Jerry.

1 0

Squeaken, anyone?
by Tim Rowledge 25 Aug '98

25 Aug '98

I've been thinking about reducing the number of computers in my life (eleven is _way_ too many) and so far it looks like it might be sensible to dump all the pda's I don't use, forget the powerbook, replace the older RPC with the new RPC2 that is on order and replace the newer RPC with a Corel NetWinder. Let's be honest, all I do with computers is mail, web & squeak, so why keep any of the others? Except for using Quicken - which is why I keep the aging Mac 8100 going. I don't think I could manage without Quicken & CheckFree. So, who out there feels like writing Squeaken. Forget all the flashy reports and graphs, just handle many accounts both asset and liability and interface to Checkfree for paying the bills. Surely someone on the list has written this before? Then I could get down to just three computers:- an RPC2, a NetWinder and a Psion5. All ARM powered, all running Squeak (well nearly, the psion work is still in slow motion mode) and no M$ anywhere. Dreaming on a hot summer's eve, tim PS Corel netwinder is _way_ cool. Look at http://www.corelcomputer.com - 275MHz ARM Linux, 64Mb ram and 4Gb for $1k. Wow. -- Useful random insult:- He's so dense, the Titanic wouldn't sink in his head. Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 842-6110 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

3 3

Re: RefactoryBrowser
by Bob Arning 25 Aug '98

25 Aug '98

On Mon, 24 Aug 1998 19:52:10 -0500 John Brant <brant(a)cs.uiuc.edu> wrote: >Do I win any prizes for guessing correctly :)... > Uh, well,,, I think the rules exclude the authors. Sorry. >There isn't a NavigatorTool in VW. Each browser contains two visual parts: >navigator and code tool. The navigator handles the selection of classes and >methods, and the code tools displays the selected item. The symbol, >#NavigatorTool, is used to name the subcanvas that will hold the navigator >part >of the browser (the navigator is determined when the browser opens). The #Tool >symbol in CodeModel is used to name the subcanvas that holds the code tool >(determined based on the navigator's selection). > Yes, I got that figured out this morning - that what looks like another Class>>#windowSpec is something inserted on the fly. They are somewhat working now. >BTW, you shouldn't need the RefactoringBrowser class to get the rewrite rule >editor working. However, if you get the navigator classes working, most of the >other interfaces are fairly simple. Well, that would depend on my knowing what I am doing. I haven't yet figured out what Mac tool will undo a .gz file so, sans documentation, it is a new mystery every minute. I started with the RRE, but if I hit the Search... button, I can wind up with a Refactory Browser which needs a #NavigatorTool. Well, that's the past - on to the future. It is pleasantly surprising how much *does* work given the time since the divergence of the Squeak and VW images. And I do look forward to actually seeing this puppy work! Cheers, Bob

2 1

Re: Block returns out of scope?
by Tim Rowledge 25 Aug '98

25 Aug '98

On Wed 26 Aug, Maurice Rabb wrote: > MessageTally>>printOn: aStream > | aSelector className | > aSelector _ class > selectorAtMethod: method > setClass: [:aClass]. <------------------- NOTE#1 > className _ aClass name contractTo: 30. > Is this a this good or bad style? How is it even possible given the > scoping once the recusive method has returned? It's not only poor style, it shouldn't compile! It only works because the compiler adds 'aClass' to the method's list of temps when it discovers it in the block, and then doesn't rescind its scope correctly and the doens't take notice of it later. I did some hacking to make block temps work a while ago that 'fixed' this, I'll try to remember to extract the relevant bits tomorrow. What should be written there is ........setClass:[:aClass| theClass := aClass]. and 'the Class' would need be a proper method temp. tim -- Useful random insult:- Barney's his hero. Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 842-6110 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

1 0

Rfeactoring Browser docs
by Bob Arning 25 Aug '98

25 Aug '98

To those who sent me the documentation for the Refactoring Browser - many thanks. To those who didn't - don't, as I have plenty to read. Cheers, Bob

1 0

Re: RV: "thisContext selector" (the revenge)
by Dan Ingalls 25 Aug '98

25 Aug '98

Ale - I'm on the road and can't send a careful answer until later. However, execute... 3 perform: #zork: with: 4 and look at the debugger state. The message is fairly nicely packaged up and will even proceed properly if you define zork: before proceeding. So I would spend a while looking at what the debugger does. It should ultimately answer your need. - Dan >what happens when you send a #perform: message? >It is solved in a primitive method that sends the message inside the VM; > activating a new method context to solve the performed message. >Then, when I look for the sent message in the context, the message found > is #perform:... and the activated method is NOT the #perform implementation >Ouch! >The #perform: implementation does not create a MethodContext because > it is a primitive method! >mmm... >When I search in the contexts (actual and the sender's context) >I can find the #perform: selector, but the arguments are missing >( has been absorbed by primitive) >Provably the problem will be difficult to solve with the contexts information only... > >Is there any known (easy) way to solve the #perform: problem? > >thanks. >Ale.

1 0

Re: True dialog windows
by Tim Olson 24 Aug '98

24 Aug '98

>Bob Arning said: >> On Sun, 23 Aug 1998 10:04:25 -0500 Tim Olson <tim(a)jumpnet.com> wrote:=20 >> >>I may be missing something, but I don't see anything that looks like = >> real=20 >> >>dialog window support in Squeak. Is there something available? >> > >> >The FillInTheBlank class provides dialog support for text entry, and=20 >> >SelectionMenu>>confirm: can be used for yes/no dialog boxes. >> > >Personally, I hate it when the system pops up a dialog and I can't use my >system to determine the answer to the question it's asking. "so, what do >you want to name that file?" (No! You can't use the system to look at what >other files you have in that directory!). I agree -- modal dialog boxes of that type are evil -- EVIL I tell 'ya! ;-) Unfortunately FillInTheBlank and SelectionMenu are modal in just this way. Much better are the type found in NeXTSTEP -- dialog boxes cause the application to wait for a response, but the box itself can be moved, and other applications and windows can be activated normally. Much like the standard User Interrupt window in Squeak. -- tim

3 2

← Newer
1
...
4
5
6
7
8
9
10
...
19
Older →

Jump to page: