Squeak-dev June 1998

squeak-dev@lists.squeakfoundation.org

83 participants
130 discussions

Re: TerseMan
by Alan C. Kay 18 Jun '98

18 Jun '98

Steve -- I would agree entirely. Cheers, Alan ---- At 2:20 PM -0700 6/17/98, Steve Dekorte wrote: >>A good way to give epiphany a chance is to contemplate the difference between a clock full of >>gears, and a living organism made from cells, which can grow itself from a single "universal" >>cell. > >The way that lead me to objects is trying to write a user interface from scratch in C. >It begins to become obvious that the widget's state information and functions should >most responably be stored together in the widget itself. > >I don't think this is as easy to see for the typical number crunching oriented >problems found in programming class projects. > >Communicating the benefit of objects may be easier by showing it's value for >high level objects first and letting people see the beauty of exting that downward >than to try to convince people straight out that "+" should be a dynamicly bound >message. > >Steve

1 0

Error
by sqrmax＠cvtci.com.ar 17 Jun '98

17 Jun '98

Hi. In my previous message, the proof is wrong because of a typo... it should go like this: Proof. For the average we need the amount of subclasses, and on the other hand the amount of classes. But each is one of the others except nil's subclasses. There's only one of those (Object), then the average of subclasses per class is (amountOfSubclasses) / amountOfClasses = (amountOfClasses + 1) /. amountOfClasses = 1 / amountOfClasses + 1, that is, arbitrarily close to 1 as the amount of classes increase. Andres.

1 0

How many subclasses does a class have?
by sqrmax＠cvtci.com.ar 17 Jun '98

17 Jun '98

Hi. data := Dictionary new. Object withAllSubclasses do: [:each | data at: each subclasses size put: (data at: each subclasses size ifAbsent: [0]) + 1] data inspect In my image this gave out: 0 subclasses: 885 1 subclass: 173 2 subclasses: 104 3: 20 4: 30 5: 8 6: 8 7: 4 8: 4 9: 8 10: 4 15: 2 16: 2 17: 2 169: 2 Now comes the funny part. What is the average amount of subclasses per class? Answer: 1 / amountOfClasses + 1. That is, almost 1. Amazing, isn't it? Proof. For the average we need the amount of subclasses, and on the other hand the amount of classes. But each is one of the others except nil's subclasses. There's only one of those (Object), then the average of subclasses per class is (amountOfSubclasses + 1) / amountOfClasses = (amountOfClasses + 1) /. amountOfClasses = 1 / amountOfClasses + 1, that is, arbitrarily close to 1 as the amount of classes increase. Andres.

4 3

squeak as a Java alternative
by jecel＠lsi.usp.br 17 Jun '98

17 Jun '98

Now that the source code for Netscape 5 is out, an interesting project would be to totally embed Squeak in it. This would be much nicer than using it as a helper application or even as a plug-in. Netscape couldn't include Java with the sources (I don't know what they have against Kaffe) but Squeak would be a perfect alternative. -- Jecel

1 0

Re: Behavior methods
by Vassili Bykov 17 Jun '98

17 Jun '98

At 12:03 AM 6/17/98 +0300, Boris G. Chr. Shingarov wrote: >(This is not exactly what is in IBM Smalltalk where allInstances >returns an OrderedCollection, contrast to Set in Squeak. I think >it should be Set there - but maybe someone knows any specific >reason for OC??) The motivation behind the choice might be the array vs. hash table implementation rather than ordered vs. unordered property of the collection. An arrayed collection is cheaper storage-wise (at least if you use it properly), and adding objects to it is faster--there is no hash code calculation, no search for a free slot in a hash bucket, and no rehashing after growing. So if a collection is created primarily for the sake of iterating over it later, rather than testing for membership using #includes:, an OC may be a better choice than a Set. (Actually, #includes: on a small OC may be faster than on a Set with same elements). --Vassili -- Vassili Bykov vassili(a)objectpeople.com The Object People http://www.objectpeople.com +1(613)225-8812 "Any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp." -- Greenspun's Tenth Rule of Programming

2 1

Twain97 support
by Alejandro F. Reimondo 16 Jun '98

16 Jun '98

Folks, I have implemented the EZTwain97 API interface and it can be used with Squeak+sqAPIs under Windows. Can be found at: http://www.sugarWeb.com/Miniatures/downloads.html cheers, Ale. Estimados, He implementado una interfaz a Twain para soportar scanners en Squeak (con sqAPI) y Visual Smalltalk tambien es util para otros perifericos twain, como camaras digitales, etc. Pueden visitar la pagina de miniaturas SUGAR. http://www.sugarWeb.com/Miniatures/Miniatures.htm saludos, Ale.

1 0

Re: Morph Must Die!!! (a bug report)
by Tim Olson 16 Jun '98

16 Jun '98

Re: breaking dependancies: Instead of reimplementing the manual release protocol, should we look into implementing WeakReferences so that the GC can automatically clean up dependants when all "strong" references have disappeared? -- tim

4 3

Behavior methods
by sqrmax＠cvtci.com.ar 16 Jun '98

16 Jun '98

Hi. I've found faster implementations for four methods of Behaviour. Maybe they're worth an update item. Two of the speedups are fourfold! Andres. Content-Type: application/octet-stream; name="Behavior updates.cs" Content-Disposition: attachment; Filename="Behavior updates.cs" Attachment converted: Cog:Behavior updates.cs (????/----) (0000D82A)

1 0

RE: Morph Must Die!!! (a bug report)
by Alejandro F. Reimondo 15 Jun '98

15 Jun '98

Hi all, Andreas Raab has implemented WeakClasses for Squeak 1.31 on 21 March 1998 Cheers, Ale. ---------- Desde: Alan C. Kay[SMTP:alank@wdi.disney.com] Enviado el: Lunes 15 de Junio de 1998 19:46 Para: squeak(a)cs.uiuc.edu Asunto: Re: Morph Must Die!!! (a bug report) Tim -- That would be cool! Do you have a first cut at a proposal? Cheers, Alan ------ At 5:34 PM -0500 6/15/98, Tim Olson wrote: >Re: breaking dependancies: > >Instead of reimplementing the manual release protocol, should we look >into implementing WeakReferences so that the GC can automatically clean >up dependants when all "strong" references have disappeared? > > > > -- tim

1 0

More care with <PRE> & </PRE> please
by Michael Irish 15 Jun '98

15 Jun '98

I have been gradually going through this archive. Every now and then I have come across a message which the browser allows to go off the right hand edge and I have to use a horizontal scroll bar. On Friday, I investigated. I think the problem is that people are using <PRE> & </PRE> inappropriately. As I understand it, those two tags are for pre-formatted text such as assembler code. If you type ordinary text between them, the browsers assume that you actually want long lines. I rather doubt that you do. Two examples that I came across on Friday and Saturday are http://squeak.cs.uiuc.edu/mail/squeak/msg00032.html and msg00117.html Michael Irish.

3 2

← Newer
1
2
3
4
5
6
7
8
9
...
13
Older →