Internet Info 1994 March/Internet Info CD-ROM (Walnut Creek) (March 1994).iso/answers/comp/smalltalk-faq

Text File | 1994-03-03 | 55.7 KB | 1,595 lines

Path: bloom-beacon.mit.edu!hookup!swrinde!ihnp4.ucsd.edu!agate!con.Berkeley.EDU!latta
From: [e-mail] (Craig R. Latta)
Newsgroups: comp.lang.smalltalk,news.answers,comp.answers
Subject: Smalltalk Frequently-Asked Questions (FAQ)
Followup-To: poster
Date: 3 Mar 1994 05:46:58 GMT
Organization: University of California, Berkeley
Lines: 1573
Approved: [e-mail]
Distribution: world
Expires: 22 Mar 1994 05:46:45 GMT
Message-ID: <[e-mail]>
Reply-To: [e-mail]
NNTP-Posting-Host: con.berkeley.edu
Summary: This posting contains a list of frequently-asked questions
        (and their answers) about the Smalltalk programming
        language and environment. It should be read by posters
        to comp.lang.smalltalk.
Originator: [e-mail]
Xref: bloom-beacon.mit.edu comp.lang.smalltalk:6580 news.answers:15967 comp.answers:4007

Archive-name: smalltalk-faq
Last-modified: 1994/2/21
Version: 3.7


***

    This is a Smalltalk frequently-asked-questions (FAQ) document,
distributed by Craig Latta <[e-mail]>. It is (ostensibly) posted
fortnightly to the USENET newsgroups comp.lang.smalltalk and
news.answers.
    It is also available via FTP as
[e-mail]:misc/smalltalk/FAQ/SmalltalkFAQ. Finally, it can
be obtained by mail by emailing Smalltalk-request@xcf with the subject
line "request for FAQ". The machine XCF.Berkeley.EDU has IP address
128.32.138.1.

    Please send contributions, suggestions and comments to
[e-mail]. They are greatly appreciated. I realize
that information has a short half-life.
    Obviously (to me, anyway), it╣s silly to force a structured document like this one to exist as a flat text stream. In the medium term, I╣m in the midst of redoing this document as hypertext (suitable for serving via the World Wide Web), and implementing a Smalltalk interface for editing and reading it in the long term. Offers of assistance are welcome.


        Thanks,

-C


***

    New items are marked with a '+'. Modified existing items are
marked with a '*'.


    Contents:

0.0)    [Meta-issues]
0.1)        How can I browse ftp sites and their data without
            using my own disk space (unless I want to keep
            data), and locate files on ftp sites, given
            pathname fragments?
0.2)*        Is comp.lang.smalltalk archived or available via
            email?

1.0)     [Archival]
1.1)         What free or low-cost Smalltalk implementations are
            there? How can I get GNU Smalltalk?
1.2)        What Smalltalk archives are there?

2.0)     [Projects]
2.1)        What is Smallmusic?
2.2)         What is CoolDraw?
2.3)+        What is SmallDraw?
2.4)+        What work is being done with neural nets in Smalltalk?

3.0)     [References]
3.1)        Can someone recommend a good introduction to
            Model-View-Controller concepts?
3.2)        Is there a Smalltalk bibliography?
3.3)        What are the "blue book", "purple book", etc?
3.4)        Who are some employers of Smalltalk programmers?
3.5)        What is the Smalltalk Report?
3.6)          Is there a GNU Smalltalk tutorial?
3.7)         What's a summary of multiple inheritance in Smalltalk?
3.8)        What's the status of the ANSI Smalltalk
            standardization effort?
3.9)        Have there been any studies done on performance
            differences between complete C++/Smalltalk
            (specifically ObjectWorks\Smalltalk)
            applications in terms of runtime and space
            consumption?
3.10)+        Where can I learn about the history of Smalltalk?
3.11)+        How can I get the Smalltalk Resource Guide?

4.0)    [Programming issues]
4.1)        What are some "classic Smalltalk bugs", both in the
            system and programmer domains?
4.2)+        How do I use Pool dictionaries?
4.3)+        When browsing sources, I see │t1▓, │t2▓, etc., instead of semantically
            relevant variable names. What╣s going on?
4.4)+        What happened to >>deepCopy in ParcPlace Smalltalk?

5.0)    ["Vendor"-specific issues]
5.1)        How compatible is GNU Smalltalk with Smalltalk-80 (or
            Smalltalk/V)?
5.2)        How do I contact the producers of various Smalltalk
            implementations?
5.3)+        How can I get information about HP distributed Smalltalk?

6.0)    [Smalltalk implementation issues]
6.1)        Is Smalltalk interpreted or compiled?

---

0.0) [Meta-issues]

---

0.1)        How can I browse ftp sites and their data without
            using my own disk space (unless I want to keep
            data), and locate files on ftp sites, given
            pathname fragments?

Answer:

    This question might seem tangential at first (and I suppose it
is). But it is vitally important, as resources such as papers,
documentation, code and software tools become more numerous and
distributed.

        There is a set of Emacs-Lisp ("elisp") code, called
"ange-ftp.el", which makes 'ftp' use transparent within GNU Emacs (GNU
Emacs is available via anonymous ftp from prep.ai.mit.edu). This
package attempts to make accessing files and directories using FTP
from within GNU Emacs as simple and transparent as possible.  A subset
of the common file-handling routines are extended to interact with
FTP. Using these routines, one is able to access remote files and one
would any other local file, without having to write it locally to
disk. The result is an immense virtual global filesystem.
        The routines are available via anonymous ftp (naturally!) as
tut.cis.ohio-state.edu:/pub/gnu/emacs/elisp-archive/as-is/ange-ftp.tar.Z,
(incidentally, if you already had "ange-ftp.el", you could paste the
above line in response to Emacs' 'copy-file', stick "/anonyous@" in
front of it, and copy the file.) My current version is dated 22
October 1991.
        Another useful bit of elisp is "saveconf.el". It saves the
Emacs buffer list and window configuration between editing sessions.
So, one can have several buffers, with several files open (as I
usually do), quit and restart Emacs, and have the state preserved,
cursor locations and windows included. Happily, it works well with
"ange-ftp.el", so that even remote files are restored (after possibly
having to prompt for passwords). "context.el" is also available via
anonymous ftp from cis.ohio-state.edu, as
pub/gnu/emacs/elisp-archive/saveconf.el.Z. Also look for
"tree-dired.el" which provides for hierarchical directory editing.
        Incidentally, it was very easy to produce references for the
above tools, thanks to another tool called "archie", developed at
McGill University. Dubbed a "resource discovery tool" by its authors,
it comes in very handy when one knows what tools are needed but not
their availability. Archie consists of a server for this information
(basically from a database of directory trees from "all known"
anonymous ftp sites, updated once per month), and a client, which may
be run via 'telnet' from the server machine itself (frowned upon...),
or from a standalone client available from that machine (...highly
encouraged, for the considerable host load win). Some clients even
perform ftp tasks based on user response to search results. There are
clients available for dumb and X terminals as well as for (of course)
Emacs. Poke around archie.mcgill.ca for a client and documentation.

    Adapting these tools (or at least new interfaces to them) for
Smalltalk would make for several great projects. I'm working on it in my spare
time. I'd love to hear from any interested people.


-Craig


---

0.2)         Is comp.lang.smalltalk archived or available via
            email?

Answer:

    Yes:

        Info-CLS (formerly Smalk) is a mailing list which is
bidirectionally gatewayed with comp.lang.smalltalk (via
[e-mail]). Every posting to c.l.s (with distribution
options USA or world) gets distributed to all subscribers of
Info-CLS, and vice versa every mailing to Info-CLS gets posted
to c.l.s.

        There is a mailarchive associated with this list, i.e.
every posting/mailing gets archived and can be keyword searched by
e-mail.  In order to search the archive, which started somewhere
around September '92, send a mail message to

   [e-mail]

containing something like:
-----------------------------------cut here----
//
Database Search DD=Rules
//Rules DD   *
Search type & checking in INFO-CLS
index
print
/*
-----------------------------------and here----
as the body part of the message. On details for more finegrained
retrievals, read the doc which is obtainable by sending
    GET LISTDB MEMO
as the body part of a mail message to the listserv.

In order to get subscribed, send a mail message to the listserv,
containing
    SUBSCRIBE Info-CLS Your I. Name
or drop me a note.

I want to add that I'm in the middle of preparing about another
years postings for upload (91-92), and am interested in further
private loggings for upload.
The archive is also available via gopher, but unfortunately only
as monthly chunks - which taaakes aaa looong tiiime to retrieve
them.  I haven't succeeded yet in convincing the staff at vm.gmd.de
to install software which can make single postings available by
subject.
On my wishlist is a system written in (GNU) Smalltalk, accessible
via TCP/IP, representing some kind of 'remote dictionary'.  But
that's far in the clouds.

-j|g (Joerg Rade [e-mail])

---

1.0)     [Archival]

---

1.1)        What free or low-cost Smalltalk implementations are
            there? How can I get GNU Smalltalk?

Answer:

    GNU Smalltalk is free. The most current location, to my
knowledge, is [e-mail]:pub/gnu/smalltalk-1.1.1.tar.Z.
Please direct problems to the author, Steven Byrne, at
[e-mail].

    Little Smalltalk, by Timothy Budd, is available as a book and
sources, at bookstores. It is also available as
[e-mail]:pub/comp.archives/little-smalltalk.

---

1.2)        What Smalltalk archives are there?

Answer:

    There are many. Most of them simply archive GNU smalltalk, but
there are also a few large archives containing many interesting and
varied sources. All of the sites may be retrieved by invoking 'archie
smalltalk' (see above reference to 'archie').
    For convenience, descriptions of a few of the archives follow.
If you have a site/announcement you'd like included, please let me
know.


**

Location: [e-mail]:misc/smalltalk
Summary:

    Smalltalk FAQ, smallmusic discussion archive.

**

Host: mushroom.cs.man.ac.uk
Summary: The Manchester Smalltalk archive. Information about it is
    posted regularly to comp.lang.smalltalk.

**

Document: [e-mail]:pub/Index
Summary: Information about the UIUC Smalltalk archive (which has local
    files and a mirror of the Manchester archive).

**

Document: [e-mail]:pub/st80/README
Summary: Information about various Smalltalk-related offerings,
    including the Musical Object Development Environment (MODE).


---

2.0)     [Projects]

---

2.1)        What is Smallmusic?

Answer:

    A work group has formed to discuss and develop an
object-oriented software system for music. The current environment is
Smalltalk 80. The email address for the group is
[e-mail].  If you are interested in joining the
discussion, email [e-mail], with the
subject line "add me".

    The abstract and outline to a recent version of our working
paper follows. The document is available via ftp as
[e-mail]:pub/st80/OOMR6.t.


    Thanks,

    Craig Latta
    [e-mail]

***

Abstract to the working document

    This document describes an object-oriented description
language for musical parameters, events and structures known as the
Smallmusic Object Kernel (SmOKe) .  In object-oriented software terms,
the representation is described in terms of software class hierarchies
of objects that share state and behavior and implement the description
language as their protocol.
    The authors believe this representation, and its proposed
linear ASCII description in Smalltalk-80 syntax, to be well-suited as
a basis for: (1) concrete description languages in other languages,
(2) specially-designed binary storage and interchange formats, and (3)
use within and between interactive multi-media, hypermedia
applications in several application domains.

    There is an article about an environment implementing SmOKe,
called the Musical Object Development Environment (MODE), in the
Computer Music Journal, volume 16 number 3. There's an "outrageously
sexy" screen shot of the MODE on the cover.

---

2.2)         What is CoolDraw?

Answer:

Newsgroups: comp.lang.smalltalk
From: [e-mail] (Bjorn Freeman-Benson)
Subject: CoolDraw - HotDraw with Constraints (or ThingLab: The Next Generation)
Organization: University of Victoria, Victoria, BC, Canada
Date: Tue, 6 Apr 93 05:15:29 GMT

A while back I saw a query in this group about updated version of ThingLab.
Well, I'm happy to announce that although the name has changed, the
philosophy is the same, and CoolDraw is now available via anonymous FTP.
Here is the READ.ME:

CoolDraw Release 3 Documentation                         5-Apr-93

I. What is CoolDraw?

CoolDraw is a constraint-based object-oriented drawing framework.  CoolDraw
is similar to other MacDraw-like packages with one major exception:
everything in CoolDraw is done with constraints.  The user interface is
done with constraints.  Constraints can be added between figures.  Even the
constraint debugger (a future feature) will be done with constraints.  The
total integration of constraints makes it "pretty darn cool".  The system
is written in ParcPlace Smalltalk-80 release 4.1.

CoolDraw is similar to other systems such as IntelliDraw, Chimera, and
Briar.  Also note that CoolDraw is a technology demonstration and research
tool -- not a fully debugged product.  Occasionally it requires some loving
care.  And, quite naturally, it has some unimplemented features.

II. How do I use it?

Open a CoolDrawingEditor using the "open" message.  On the left is a tool
palette.  The top tools are the usual drawing tools: selection, move to
front, move to back, erase, etc.  The middle tools are figure creation
tools: line, rectangle, ellipse, and so on.  The bottom four tools are the
constraint creation and deletion tools: three for creation (one variable,
two variable, and three variable constraints) and one for deletion.

While you automatically use constraints by just creating figures and moving
them around on the screen (because the entire system uses constraints
internally), if you want to create explicit constraints, you have to use
one of the tools.  The cursor of the tool changes to indicate how many
figures have been selected and when enough are selected a dialog box
appears with a standard selection of constraints.  If you want some other
constraint, then press the "Custom" button and a second dialog box offering
a complete selection of constraints will appear.

When the constraint deletion tool is clicked on a figure, it offers a list
of all the explicit constraints attached to that figure.  Any number can be
selected and deleted at once.

III. How do I get it?

Anonymous FTP from ursamajor.uvic.ca in ~ftp/ursa/constraints/CoolDraw
There are two options.  Get the image:
    cooldraw.image.tar.Z
or get the source code:
    cooldraw.source.tar.Z

IV. What are the bugs?

Numerous.  The major flaws are that the constraint system is not currently
powerful enough to solve cycles or simultaneous equations, and that there
is not a large selection of figures.  Minor ones include that the three
constraint tool is not implemented.

The most noticable un-bug is that the constraints do not always do what you
expect.  This is a well-known problem is constraint research and has no
known solution short of a mind-reading computer.  The constraint
hierarchies that CoolDraw uses can help, and future versions will have more
default weak constraints to create "expected" behavior.

V. What are the future plans?

We intend to continue work on three fronts: (1) fix the details (make more
figures available; fix the dialog boxes; provide a wider assortment of
built-in constraints; implement the three constraint tool; etc.); (2)
improve the constraint solver to deal with more complex constraints yet
retain its efficiency; (3) add other cool things such as constraint
debugging tools, animation, etc.

VI. What else should I know?

Without explicitly inspecting the drawing, there is no access to the
internal consistency constraints of the figures.

VIII. Credits and Copyright

See the startup screen.

The CoolDraw code is copyright 1993 by Bjorn N. Freeman-Benson, and its
commercial use is restricted.  For information about the HotDraw or SkyBlue
code, contact the respective authors: Ralph Johnson and Michael Sannella.

Naturally, we would be happy to hear from anyone who uses, plays around
with, or just looks at the system.

Regards,
Bjorn N. Freeman-Benson

---

2.3)        What is SmallDraw?

From: [e-mail] (Dan Benson)
Subject: SmallDraw for VisualWorks
Organization: Siemens Corporate Research, Inc
Date: Tue, 29 Jun 1993 18:37:28 GMT

I recently submitted a version of SmallDraw that now runs under
VisualWorks to the Illinois archive. It's availabe via anonymous
ftp at: st.cs.uiuc.edu as /pub/st80_vw/SmallDraw.VW

SmallDraw is a very simple structured graphics editor that was
originally written to run under Release 4.0. I wrote a series
of three articles for The Smalltalk Report last year in which
I used the SmallDraw application to illustrate Release 4 graphics
and MVC construction. The modifications to SmallDraw to make
it run under VisualWorks were made by David Arctur
([e-mail]). Thank you David!


Dan Benson                 Siemens Corporate Research, Inc.
[phone]             755 College Road East
[e-mail]    Princeton, NJ  08540

---

2.4)+        What work is being done with neural nets in Smalltalk?

From: [e-mail] ( GF)
Newsgroups: comp.ai.neural-nets,comp.lang.smalltalk
Date: 3 Nov 93 10:25:15 GMT
Organization: TOMCAT, Germering, Germany

There exists one small implementation of
"The simple Neural Network for taught Samples Distinguishing", which has been
realised at Technical University in Prague (CVUT) supported by Tomcat.
This first version is implemented in ST/V, and will be ported to ST/X in
the next future.

For more information the best way is to contact us (with your address),
we will send you some papers...,
or to contact the CVUT directly via eMail:
[e-mail] Dr. Mueller
[e-mail] Dr. Kolar
(both are Docs at the university and co-owners of FELTECH ltd.
specialised in OO-Technology)

Richard.

Tomcat computer GmbH
Richard v. Lavante
Fruehlingstr. 4
D-82110 Germering bei Muenchen
Germany
eMail [e-mail]
Tel. [phone]
Fax. [phone]

---

3.0)     [References]

---

3.1)        Can someone recommend a good introduction to
            Model-View-Controller concepts?

Answer:

From: [e-mail] (Dr. Raimund K. Ege)
Newsgroups: comp.lang.smalltalk
Subject: Re: MVC -- good introductions?
Date: 8 Mar 92 18:26:40 GMT
Organization: Florida International Univ.

Look at Chapter 10 in the following book that just came out:

Programming in an Object-Oriented Environment,
by Raimund K. Ege

Academic Press, Inc., San Diego, CA, 1992, hardcover,
ISBN 0-12-232930-9

To order call 1-800-321-5068.

(also: Academic Press Limited, London, United Kingdom)

It presents a complete and thorough introduction to all object-oriented
concepts. It contains a large
example/case study, and a comparison of major OO programming languages.

In addition, the book extends the object-oriented view
to all elements of the programming environment: data structures
and algorithms, programming tools, user interfaces, data bases and
software design.

Chapter 10 is on user interfaces: it describes and illustrates
the Smalltalk MVC paradigm (also: InterViews)

--
Raimund K. Ege                             School of Computer Science
                                             Florida Int'l University
[e-mail]           [phone]              University Park
[e-mail]     FAX [phone]              Miami, FL 33199

**

From: [e-mail] (Nils Erik Asmundvaag)
Newsgroups: comp.lang.smalltalk
Subject: Re: MVC -- good introductions?
Date: 11 Mar 92 10:56:38 GMT
Organization: Glasgow University Computing Science Dept.

The book

    Smalltalk-80: A Practical Introduction
    (ISBN 0-273-03105-8)

    by Philip D. Gray & Ramzan Mohamed, 1990
    and published by Pitman (at least in the UK)

contains two chapters on interactive applications and the MVC.

I found it very helpful when first learning about the MVC.

Nils E. Asmundvaag

--
-------------------------------------------------------------------------------
Nils Erik Asmundvaag
University of Glasgow, Scotland
[e-mail]      [e-mail]


**

From: [e-mail] (Bruce Samuelson)
Newsgroups: comp.lang.smalltalk
Subject: Re: how are st80 views and controllers used?
Date: 12 Mar 92 14:12:48 GMT
Organization: UTexas at Arlington, Linguistics

There are two papers on MVC that provide an introductory overview of
the pre-version 4.0 ST80 scheme.  Much of what's in them also applies
to version 4.0.  I understand that one of PPS's priorities in the
forthcoming version 4.1 will be improved documentation.  We'll see how
well they explain the new windowing scheme launched in version 4.0.
It would certainly be helpful to get an overview of what's going on
before plunging into the source code of the myriad new classes.

(1) A Cookbook for using the Model-View-Controller User Interface
Paradigm in Smalltalk-80 by Glenn E. Krasner and Stephen T. Pope,
Journal of Object-Oriented Programming, volume 1 number 3, pp. 26-49,
1988

(2) Applications Programming in Smalltalk-80: How to Use
Model-View-Controller (MVC) by Steve Burbeck, Softsmarts, Inc.,
copyright 1987.

The second paper is probably no longer available.  I think Softsmarts
is the company that used to sell a version of Smalltalk for 80286
machines but went out of business some years ago.  The phone number on
the paper is listed as 415-327-8100 (Palo Alto, California).  You may
try asking ParcPlace (or, less likely, Digitalk) if they have copies
of Burbeck's paper.

--
**********************************************************
* Bruce Samuelson    Department of Linguistics     *
* [e-mail]    University of Texas at Arlington *
**********************************************************


---

3.2)        Is there a Smalltalk bibliography?

Answer:

    There are many... here is one:

From: [e-mail] (Ronald Schultz)
Newsgroups: comp.lang.smalltalk
Subject: Smalltalk Relevant Texts
Date: 10 Jan 92 16:08:05 GMT
Organization: Grebyn Timesharing


A list of Smalltalk-relevant texts.  Retrieved from the Digitalk
forum on Compuserve.  If you know of any additional texts, please
let me know.  Thanx.


==========================================================================
Ron Schultz
Berard Software Engineering, Inc.
Columbus Ohio Office                           Headquarters
5634 Claire Court                              301 Lakeforest Drive
Dublin, Ohio 43017                             Gaithersburg, Md. 20877
Phone  [phone]                          [phone]
FAX    [phone]                          [phone]
=========================================================================

Smalltalk 80 The Language, Adele Goldberg & David Robson
        Addison-Wesley 1989             ISBN 0-201-13688-0

Smalltalk 80 The Interactive Programming Environment, Adele Goldberg
        Addison Wesley 1984             ISBN 0-201-11372-4

Smalltalk 80 Bits of History, Words of Advice , Glenn Krasner
        Addison Wesley 1984             ISBN 0-201-11669-3

Inside Smalltalk Volume I, Wilf Lalonde & John Pugh
        Prentice Hall 1991              ISBN 0-13-468414-1

Inside Smalltalk Volume II, Wilf Lalonde & John Pugh
        Prentice Hall 1991              ISBN 0-13-465964-3

Object-Oriented Graphics, P. Wisskirchen
    Springer-Verlag 1990        ISBN 3-540-52859-8

Practical Smalltalk: Using Smalltalk/V, Dan Shafer and Dean A. Ritz.
        Springer-Verlag                 ISBN 0-387-97394-X

Rapid Prototyping for Object Oriented Systems, Mark Mullen
        Addison Wesley 1990             ISBN 0-201-55024-5

Object-Oriented Design, Peter Coad and Ed Yourdon
    Yourdon Press 1991        ISBN 0-13-630070-7

Object Oriented Programming for Artificial Intelligence, Ernest Tello
        Addison Wesley 1989             ISBN 0-201-09228-x

The Well Tempered Object, Stephen Travis Pope
        MIT Press 1991                  ISBN 0-262-16126-5

RefTalk/Vwin, David Carl O'Neal
        NuVista Press 1991              ISBN pending

Human-Computer Interface Design Guidelines, C. Marlin Brown
        Ablex Publishing 1989           ISBN 0-89391-332-4

Designing Object-Oriented Software,
                Rebecca Wirfs-Brock, Brian Wilkerson, and Lauren Wiener
        Prentice-Hall 1990              ISBN 0-13-629825-7

Object Oriented Programming with Smalltalk/V, Dusko Savic
        Ellis Horwood 1990              ISBN 0-13-040692-9

An Introduction to Object Oriented Programming & Smalltalk
        Lewis Pinson & Richard Wiener
        Addison Wesley 1988             ISBN 0-201-19127-x

SAA Common User Access Advanced Interface Design Guide
        IBM 1989                        IBM Document # SC26-4582-0

IBM Red Books----(available from your IBM representative
                  contact your local office of IBM and request
                  the placing of an IBM Red Book Order.  If you
                  are an IBM customer, the books are free.  If you
                  are not an IBM customer, the books may have a nominal
                  fee.)----

A Practical Introduction to Object Oriented Programming
        IBM 1990                        IBM Document # GG24-3641

Object Oriented Design - A preliminary Approach
        IBM 1990                        IBM Document # GG24-3647

Developing a CUA Workplace Application
        IBM 1990                        IBM Document # GG24-3580-00

Managing the Development of Object Oriented Applications
        IBM 1990                        IBM Document # GG24-3581-00

Object Oriented Analysis of the ITSO Common Scenario
        IBM 1990                        IBM Document # GG24-3566

CUA Evaluation
        IBM 1990                        IBM Document # GG24-3456

SAA CUA '91 Guide
        IBM 1991                        IBM Document # SC34-4289

SAA CUA '91 Reference
        IBM 1991                        IBM Document # SC34-4290

SAA - A Guide for Evaluating Applications
        IBM 1991                        IBM Document # G320-9803


---

3.3)         What are the "blue book", "purple book", etc?

Answer:

Date: Wed, 11 Nov 92 12:52:39 PST
From: [e-mail] (Mike Khaw)


blue
        Goldberg, Adele, and David Robson, _Smalltalk-80: The Language
        and Its Implementation_, Addison-Wesley, 1983. ISBN
    0-201-11371-6. *Out of print*

orange
        Goldberg, Adele, _Smalltalk-80: the Interactive Programming
        Environment_, Addison-Wesley, 1984. ISBN 0-201-11372-4.

green
        Krasner, Glenn, ed., _Smalltalk-80: Bits of History, Words of
        Advice_, Addison-Wesley, 1983, ISBN 0-201-11669-3

purple
        Goldberg, Adele, and David Robson, _Smalltalk-80: The Language_,
        Addison-Wesley, 1989, ISBN 0-201-13688-0

The books are actually cream or tan. The color referred to is the color
used as the background of the illustration on the front cover (as well
as for the Addison-Wesley logo on the spine).

The purple book is an update/revision of the blue book, with the
section on the abstract bytecode machine omitted (because it was out
of date, according to Adele).
----------

Mike


---

3.4)         Who are some employers of Smalltalk programmers?

Answer:

From: [e-mail] (Ralph Johnson)
Newsgroups: comp.lang.smalltalk
Subject: Smalltalk employers file
Date: 18 Sep 92 16:32:43 GMT
Organization: University of Illinois, Dept. of Comp. Sci., Urbana, IL

    I've noticed a rash of job offers here (I approve of them) and
decided that I should distribute my list of companies hiring
Smalltalkers.  I have collected this over the years and give it to my
students.  However, the size of the list is significantly greater than
the number of my students, so I decided to make it public. I would be
happy to add entries, remove entries, or make other changes to it.  If
you are an employer then you might want to take a look at it to see
whether you like what I say about you, and let me know if you don't.
I'll change it, though I like *short* entries and have shortened
almost everything in here.

    The file is in the Smalltalk archive in
/st-docs/smalltalk-jobs.  You can ftp it from st.cs.uiuc.edu or you
can use the e-mail archive server (see the recent FAQ).  Europeans
note that this will automatically show up in the Manchester archive in
a day or so.

-Ralph Johnson

---

3.5)        What is the Smalltalk Report?

From: [e-mail] (Markus Stumptner)
Newsgroups: comp.lang.smalltalk
Subject: Opinions on Smalltalk Report (Summary)
Date: 12 Feb 93 11:34:32 GMT
Organization: DB and ES Subdivision, TU Vienna

Ok, here are the comments I received about the Smalltalk Report.
While the sample base is quite small, the response seems
overwhelmingly favorable.  Thanks to all who responded.

    Markus

---------------------------------------------------------------------
I have subscribed since the first issue and I find it to be an excellent
combination of theoretical and practical articles. John Pugh (of LaLonde
and Pugh) is the editor.

On the other hand, I will warn you to expect subscription headaches with
SIGS Publications. They are probably the worst I've ever dealt with and
our technical librarian agrees. They are barely capable of processing a
renewal let alone a new subscription or problem.
---------------------------------------------------------------------
We subscribe, last issue was about 20 pages, 2 articles, 3 columns.
I would call it a must read publication.
---------------------------------------------------------------------
I think its pretty good. (But then I should mention, I write for it!) It tends
toward intermediate, with lots of practical tips and examples.
---------------------------------------------------------------------
(This was not an answer to my question, but I include it for completeness.)
It is a journal published 9 times per year. Cost in US is $69. Editor
with whom I deal is Paul White ([e-mail]). The Smalltalk
Report, Subscribers Services, Dept. SML, PO Box 3000, Denville, NJ
07834-9821.
---------------------------------------------------------------------


--
Markus Stumptner                                [e-mail]
University of Technology Vienna                 vexpert![e-mail]
Paniglg. 16, A-1040 Vienna, Austria             ...mcsun!vexpert!mst


Newsgroups: comp.lang.smalltalk
From: [e-mail] (Alan Knight)
Subject: Re: Smalltalk Report
Organization: Carleton University
Date: Mon, 15 Feb 1993 22:30:55 GMT

The Smalltalk report
Subscriber Services Dept SML
PO Box 3000
Denville NJ USA
07834-9821

FAX [phone]

ISSN 1056-7976

US$69/year in e US
US$94/year outside the US

---

3.6)          Is there a GNU Smalltalk tutorial?

Answer:

    Yes.

From: [e-mail] (Steve Byrne)
Date: 28 Feb 1993 03:00:10 GMT
Organization: Sun Microsystems Inc., Mountain View, CA


    Andy Valencia has very graciously provided an extended
tutorial document describing the Smalltalk language with respect to
GNU Smalltalk.  It covers a wide variety of topics, and provides a
general introduction to object oriented programming concepts.

    You may FTP a copy from the usual GNU locations in a few days
(after it has propagated).  Currently, it may be found at:

    prep.ai.mit.edu: pub/gnu/smalltalk-tutorial.ps.gz  (PostScript)

-or-

    prep.ai.mit.edu: pub/gnu/smalltalk-tutorial.txt.gz  (ASCII)

Check it out -- this looks really good!

Steve

--

3.7)         What's a summary of multiple inheritance in Smalltalk?

Answer:

From: [e-mail] (Mario Wolczko)
Newsgroups: comp.lang.smalltalk
Subject: Re: Multiple Inheritance in Smalltalk??
Date: 20 Apr 93 12:40:27 GMT
Organization: Dept Computer Science, University of Manchester, U.K.

In article <[e-mail]>, [e-mail] (cs331106) writes:
>
>   I have been researching the smalltalk language, and have found
> conflicting reports as to whether or not smalltalk is a single
> or multiple inheritance style language.  Any answer would be
> appreciated!

Yes and no, mostly no.  The initial versions of Smalltalk did not have
multiple inheritance.  An MI scheme, written entirely in Smalltalk (ie
not requiring any virtual machine support), was subsequently added.
It is described in
@inproceedings{BorningIngalls82:multipleinheritance,
    author =    "A. H. Borning and D. H. H. Ingalls",
    address =    "Pittsburgh, PA",
    year =        1982,
    booktitle =    "Proceedings of National Conference on
    Artificial Intelligence",
    pages =        "234-237",
    title =        "Multiple inheritance in {S}malltalk-80"
}

This was shipped as part of the Smalltalk-80 virtual image, at least
from version 2.0 (the first "public" release, I believe), through 2.3.
It was removed in 2.4 (or possibly 2.5).  There were several
technical problems with the implementation, and it didn't look like
anyone was really using it anyway.  It was never thoroughly integrated
into the system (eg browsers and other tools).

Mario Wolczko
   ______      Dept. of Computer Science   Internet:      [e-mail]
 /~      ~\    The University              uucp:    mcsun!uknet!man.cs!mario
(    __    )   Manchester M13 9PL          JANET:         [e-mail]
 `-':  :`-'    U.K.                        Tel: [phone]  (FAX: 6236)
____;  ;_____________the mushroom project___________________________________


---

3.8)         What's the status of the ANSI Smalltalk
            standardization effort?

Answer:

    [Here are the minutes from the the first ANSI Smalltalk
meeting. Unless available elsewise, I'll archive future ANSI Smalltalk
materials I come across on XCF.Berkeley.EDU, separately.  --crl]

From: [e-mail] (Rick DeNatale)
Date: Tue, 6 Jul 93 15:17:30 EDT
Newsgroups: comp.lang.smalltalk
Subject: Minutes of X3J20, Smalltalk


X3J20, Programming Language: Smalltalk held its organizational
meeting at CBEMA headquarters, Washington D.C. on June 21, 1993.

1.0 The meeting was convened at 10:00 a.m. Yen-Ping Shan of IBM
acted as meeting chairman.

2.0 Attending were (listed in alphabetical order by Company):

   Aladdin Enterprises
     L. Peter Deutsch    email: [e-mail]
   Andersen Consulting
     Roy Underwood              [e-mail]
     Chris Frye (Alternate)
   Beckman Software Engineering
     Mel Beckman                [e-mail]
   Digitalk
     Allen Wirfs-Brock          [e-mail]
     George Bosworth (Alternate) [e-mail]
   Easel Corporation
     Chris Norman          [e-mail]
   International Business Machines Corporation
     Yen-Ping Shan          [e-mail]
     Rick DeNatale (Alternate)  [e-mail]
   Knowledge Systems Corporation
     Jay Casler
   Object Technology International
     Brian Barry              [e-mail]
   ParcPlace Systems
     Glenn Krasner            [e-mail]
     Richard Dym (Alternate)
   Pillar Systems
     Seyho Chang             [e-mail]
     Patti Dock (Alternate)  [e-mail]
   Quasar Knowledge Systems
     David Simmons           [e-mail]
   Servio Corp.
     Bruce Schuchardt        [e-mail]
   U.S. Geological Survey
     Jim Fulton           [e-mail]

Bob Follett Chairman of JTC1/SC22 attended the meeting as an
observer and guide to X3.

3.0 Agenda
Peter Deutsch wished to add a discussion of a proposed language standard
which was developed in 1987 in addition to the scheduled discussion of
the Common Base document (X3J20/93-003).  Peter submitted a copy of this
document which was copied and distributed as document X3J20/93-004.
Peter then said that there were many issues to deal with in setting
the goals and scope of a Smalltalk standard including: level of
formality; relationship to existing/proposed implementations;
coverage; standardization time-frame; consolidation vs. innovation.

It was generally agreed to add a discussion of goals before discussion
of proposals for a base document.

The resulting agenda is given at the end of these minutes.

Jay Casler moved to approve the agenda as modified, Dave Simmons
seconded and it was unanimously agreed.

4.0 Kate McMillan then gave an X3 overview.

5.0 Review of Project Proposal

The general discussion turned around what level was appropriate for
a Smalltalk Standard.  There was a discussion of the various
potential business and technical goals, which vary for users and
implementors of the language. Various positions of conservatism
vs. radicalism are possible with regards to many issues such as
multiple inheritance, scoping of block arguments or concurrency.

There is a trade-off between the speed of standardization and
the scope of the standard.

It was decided that members would prepare position papers on
the scope and goals for X3J20 which would cover issues including:

    Technical and Business goals
    Project Scope
    Proposals for detailed work plan
    Dimensions to be considered, e.g.:
       Methods/rigor of specification
       Implementations vs. Language Features
       Can the Standard be factored into parts,and what are
         the dependencies.

This was accepted without objection.

Mel Beckman volunteered to act as a consolidator and to prepare
a report on the positions to serve as a focus for discussion at the
next meeting.

Bruce Schuchardt, volunteered to set up a mail reflector so that
everyone will see position papers as they are developed and can
comment on the positions.  Members are encouraged to develop their
positions interactively (sort of a delphi process).

Mel will base his consolidation on positions submitted by
August 15, 1993.  The consolidation report will be distributed by
October 5th 1993.

6.0 Organizational Matters

It was agreed that, pending the result of the call for volunteers,
Yen-Ping Shan would continue as acting chairman. Glenn Krasner
volunteered as acting vice-chairman, and Rick DeNatale volunteered
as acting Secretary.

7.0 Technical Discussion and Development of Work Plan

It was unanimously agreed to defer agenda item 7 until the review
of positions on the goal and scope of X3J20 has been completed.

8.0 Meeting Schedule

The following schedule for future meetings was agreed:

   October 19-21 Sunnyvale, California, host ParcPlace.
   February 1994 RTP NC, hosts: IBM and Knowledge Systems
   June 1994, Portland Or. host Digitalk
   October 1994, Chicago Ill. host Arthur Anderson


ParcPlace agreed to make machines available at the next meeting so
that different Smalltalk implementations can be viewed. Glenn Krasner
will be the contact for machine requirements.

9.0 Review of Action Items

10.0 Adjournment

Peter Deutsch moved to adjourn the meeting. It was passed without
objection and the meeting adjourned at approximately 4:00 p.m.

Approved Agenda
  1. Opening
  2. Introductions
  3. Approval of the agenda (X3J20/93-001)
  4. X3 Overview
  5. Review of Project Proposal (X3J20/93-0015R)
       Discussion of Goals and Scope
  6. Organizational Matters
  7. Technical Discussion and development of work plan
     7.1 Proposed 1987 Smalltalk Language Specification (X3J20/93-004)
     7.2 Common Base Specification (X3J20/93-003)
     7.3 Time table for milestones
  8. Future Meetings (schedule and hosts)
  9. Review of action items
 10. Adjournment


Rick DeNatale

---

3.9)+        Have there been any studies done on performance
            differences between complete C++/Smalltalk
            (specifically ObjectWorks\Smalltalk)
            applications in terms of runtime and space
            consumption?

Answer:

    See [e-mail]:misc/smalltalk/performance.

---

3.10)+        Where can I learn about the history of Smalltalk?

From: [e-mail] (Jecel Mattos de Assumpcao Jr. (kofuji))
Newsgroups: comp.lang.smalltalk
Subject: Re: History of Smalltalk
Date: 16 Dec 1993 23:31:22 GMT
Organization: Laboratorio de Sistemas Integraveis - USP - Brazil
Reply-To: [e-mail]

[The History of Programming Languages Conferences are] the best place to find out about Smalltalk history. Here is a more specific reference:
--------------------------------------------------------
"The Early History of Smalltalk"
Alan C. Kay
The Second ACM SIGPLAN History of Programming
               Languages Conference (HOPL-II)
ACM SIGPLAN Notices Vol 28 Number 3 March 1993 pp. 69-75
---------------------------------------------------------

[That issue contains preprints from the conference. It is indeed very good. In it, the editors mention that a book from the conference is forthcoming. It may be out; I haven╣t heard.  --crl]

> > 8) I've never seen the famous article within the Byte (August-1980) --
> >    it was impossible to get information about this article or even the
> >    article itself. What was the content of this article? It's christmas
> >    soon: I *really* would like to get a copy of this article :-)
>
> The entire ISSUE was on Smalltalk.  Do you mean to say that none
> of your CS libraries have August 80 Byte?  Hmm, I'll have to guard
> mine very carefully.   And I'll see what I can do to brighten your
> Christmas!

Actually, the August 80 Byte was about Forth. But the August 1981 Byte
has been out of print for many years, and is about the only back issue
you can't get. They should reprint those articles as a book. I've just
lent mine, so this is from memory:

 - there are some introductory articles and some on the philosophy of
   the system and object oriented programming.
 - there are articles about the graphical programming environment
   ( first look at cut/copy/paste for most people ), the graphic
   prmitives ( first look at bitblt ) and drawing programs.
 - the implementation is explained, as is an older version of
   virtual memory.
 - there is an excelent article on creating data structures in
   Smalltalk, as well as one on creating control structures.

Jecel

---

3.11)+        How can I get the Smalltalk Resource Guide?

Date: Mon, 10 Jan 1994 14:48:58 -0600
From: [e-mail]
Subject: smalltalk resource guide

>Has anyone seen a publication
>_Smalltalk Resource Guide_ ?

It is published and distributed by
        Creative Digital Solutions
        293 Corbett Avenue
        San Francisco, CS 94114
        415.621.4252 (fax and telephone)
        [e-mail]
        [e-mail]

The price is $45, and if you use Smalltalk professionally
(i.e. if it is worth money to you to know about Smalltalk)
then you should get it.  I was surprised to see so many
packages I didn't know about in it.

As they say about themselves:
        The Smalltalk Resource Guide is the only Smalltalk-specific
        directory o Smalltalk products and services.  it includes
        products, third-party tools, user groups, electronic
        resources, distributors, publications, trade shows,
        professional organizations, and more.  It is updated
        as necessary.

-Ralph Johnson

---

4.0)     [Programming issues]

---

4.1)         What are some "classic Smalltalk bugs", both in the
            system and programmer domains?

Answer:

    See [e-mail]:pub/st-docs/classic-bugs, and the
equivalent place on the Manchester archive (see question 1.2 above for
details).

---

4.2)+        How do I use Pool dictionaries?

    [Following is a │tutorial▓ of sorts about Pool dictionaries. --crl]

Newsgroups: comp.lang.smalltalk
From: [e-mail] (Jan Steinman, Bytesmiths)
Organization: Bytesmiths, a technical services company
Date: Thu, 22 Jul 1993 10:51:14 GMT

    [Someone] writes:

>Firstly I just can't seem to figure out how to define "Pool Dictionaries".

They must exist as a global Dictionary first.
        (Smalltalk includesKey: #MyNewPool) ifFalse:
                [Smalltalk at: #MyNewPool put: Dictionary new]

Since the individual pool variables are actually stored as Associations
in the using method's literal array (this is important to remember!),
using a Dictionary instead of an IdentityDictionary is needed here.
Note the existance test -- if you simply create a new pool Dictionary
when one already exists, all the methods that refer to pool variables
will continue to refer to the old ones. This causes no end of
confusion!

Then plop 'MyNewPool' into your class definition.

>Also my understanding of Pool Variables is that they have to be "constant"
>objects, i.e. objects which do not change values, and you just want to share
>them among objects without having to explicitly pass them.

Not exactly. As I mentioned, the pool Dictionary Association is what is
held. This means that you can change the value of a Pool Variable at
will, with any object, as long as you don't replace the Association in
the global Dictionary.

One source of problems is to declare a pool
global as an IdentityDictionary, which doesn't actually store
Associations; it makes them up on the fly whenever asked for one. In
this case, each user of a pool variable will have a different copy, and
changing one will not change the others.

Another common problem is forward declarations and file-in. Your global
Dictionary and all the keys you will use *must* be declared prior to
file-in, or else the filed-in code will refer to Associons in
Undeclared, rather than in your global Dictionary. It's possible to
swizzle the Association from Undeclared to the proper pool, but it's
better to just make sure it exists first.

>I have an object called "Blackbaord", and I want other objects in my system to have
>access to its current state. These objects not only acess the Blackboard but
>also send messages to update the blackboard.

This sounds like a simple global to me. Two reasons why you might
prefer pools:

1) you want to limit their scope to a defined set of unrelated classes
(and their subclasses), or

2) you have more than one object to share, and you don't want to
clutter the global name space.

Pool variables provide an efficient solution to a whole class of
problems, but they are little understood, and poorly supported. The
compiler could do a lot of the monkey-business that programmers
presently have to manage when they use pool variables.

 Jan Steinman, Bytesmiths
 2002 Parkside Court, West Linn, OR 97068-2767 USA, [phone]
 Friedlistrasse 19, CH-3006, Bern, Switzerland, [phone]

---

4.3)+        When browsing sources, I see │t1▓, │t2▓, etc., instead of semantically
            relevant variable names. What╣s going on?

From: Arno Breunese <[e-mail]>

On Oct 4th, [someone] said:

> Before using VW-1.0a/OW-4.1 on Solaris, I tried an old version of OW (2.5) on
> Mac. One thing I did REALLY appreciate was the self documenting environment.
> Indeed almost each class and method from the standard ST environment was
> commented and all formal arguments had clear names. Also, the explain
> command was very helpful. On Sun, I had the nasty surprise to see that formal
> parameters are now all named t1, t2, t3, ... and that there are no more comments.
> This is very bad news!! So my question is: did I miss some crucial point when
> installing VW/OW or is it just a new product policy from ParcPlace? I would be
> very interested to hear from other people having the same problem and from
> someone working at PP.

I'm sorry to say, but it's your own fault. And you are not alone in making this
mistake (see previous postings to this newsgroup). What happens is this:

 In ParcPlace Smalltalk (OW/VW), the source of all code in the image is
kept in two files: the sources file (containing the code of the basic
classes provided with the installation) and the changes file (with all
the changes and additions you made to the originally installed
sources).
When you are browsing code, the System Browser (or anything else you
use to browse) looks in the global variable SourceFiles to find the
sources and changes files. If these are present, you will see the code
as it was originally entered, with nice argument names, with comments
etc.
If the entries in SourceFiles refer to non-existent files, or if they
are nil, then the code is reconstructed from the bytecode. Obviously,
no names and no comments can be reconstructed. That's were the t1, t2,
... come from.

What you should do is open a System Workspace (which contains hints to
change the SourceFiles global) or open an Inspector on SourceFiles
rightaway. Then make sure that these variables point to existing files.
You have to take care that the set of sources file, changes file and
image file are belonging together. Using a changes file from a
different project leads to _really_strange_ effects!

Arno Breunese,
Control Laboratory,
Department of Electrical Engineering,
University of Twente,
PO Box 217,
7500 AE  Enschede,
The Netherlands

e-mail  [e-mail]

---

4.4)+        What happened to >>deepCopy in ParcPlace Smalltalk?

Newsgroups: comp.lang.smalltalk
From: [e-mail] (John Woods)
Subject: Re: deepCopy in St80 4.1
Organization: Edinburgh University
Date: Fri, 3 Dec 1993 10:55:31 GMT

[e-mail] (Rajesh Piplani) writes:

>To all the smalltalk-80 gurus out there:
>I need to make deep copy of an object. The book on smalltalk-80
>(The language) talks about the existence of this method
>(deepCopy), but my installation on unix (release 4.1) does not
>have it. Could somebody help me out? I basically need to copy an
>object with its own instance variables.
>Raj

deepCopy was removed from 4.1 because of problems (cyclic dependencies and
other things broke it). You can file it in from the backward compatability files
in the distribution.  Or you can implement your own method --- a couple of
good-looking ones have been posted here recently.  Check the archives.

                                ... John Woods

**

Newsgroups: comp.lang.smalltalk
From: [e-mail] (Dave Robbins)
Subject: Re: deepCopy in St80 4.1
Reply-To: [e-mail]
Organization: GTE Laboratories Incorporated
Date: Fri, 3 Dec 1993 21:01:52 GMT


I would suggest implementing a postCopy method in the class whose instance
variables need to be copied.  Object>shallowCopy invokes postCopy after
completing the basic copying; code in postCopy can then make copies of the
instance variables:

   postCopy
      super postCopy.
      instanceVariable1 := instanceVariable1 copy.
      instanceVariable2 := instanceVariable2 copy

and so on.  Browse all implementers of postCopy to see examples of how this
is done in quite a few classes in the 4.1 image.

The 4.1 Release Notes explain (pages 13 and 34) that deepCopy was removed,
and why.  In particular, note on page 34 the statement that deepCopy won't
even be available as backward-compatibility code in the successor to 4.1.

Dave Robbins                    GTE Laboratories Incorporated
[e-mail]                40 Sylvan Rd.
...!harvard!bunny!drobbins      Waltham, MA 02254

---

5.0)     ["Vendor"-specific issues]

---

5.1)         How compatible is GNU Smalltalk with Smalltalk-80 (or
            SmalltalkV)?

From: [e-mail] (Steve Byrne)
Newsgroups: comp.lang.smalltalk
Subject: Re: GNU Smalltalk Compatibility
Date: 12 Dec 92 00:52:04
Organization: FSF hackers, Smalltalk division

In article <1992Dec2.182201.11980@ntmtv> [e-mail] (Chuck
Irvine) writes:

   How compatible is GNU Smalltalk with Smalltalk-80 (or SmalltalkV)?
   Just curious. Thanks

    This is probably in the FAQ list [it is now...  --crl], but
here goes anyway.  GNU Smalltalk tries to be syntax compatible with
the language described in the blue/purple book (ST-80: the Language*).
I think the only discrepancy is that GNU Smalltalk doesn't allow block
local temporary variables.

Steve

    [Note, however, that the class library for GNU Smalltalk is
very much in its infancy. Most importantly, it has little in the way
of graphical user-interface classes.  --crl]

---

5.2)*        How do I contact producers of various Smalltalk
            implementations?

Answer:

Newsgroups: comp.lang.smalltalk
From: [e-mail] (Martin Brown)
Subject: FYI: new Digitalk addresses
Organization: NETCOM On-line Communication Services (408 241-9760 guest)
Date: Tue, 21 Sep 1993 09:10:44 GMT

    Digitalk:

Effective Monday, September 20, all Digitalk departments EXCEPT development,
support and documentation have moved to brand-new, shining corporate offices
in Orange County, CA.  The new address and phone:

  Digitalk, Inc.
  5 Hutton Center
  Santa Ana, CA   92707-5754
  714-513-3000 (phone)
  714-513-3100 (fax)

Digitalk also has offices in Portland, Chicago and New York.  The numbers are:

  Digitalk Portland (training, consulting)
  503-691-0800 (phone)
  503-691-2742 (fax)

  Digitalk Chicago (sales)
  312-444-2065 (phone)
  312-641-3096 (fax)

  Digitalk New York (sales)
  212-808-3068 (phone)
  212-808-3020 (fax)

[message ends]

    ParcPlace:

    ParcPlace information may be reached at [e-mail].

    [feel free to contribute additional info  --crl]

---

5.3)+        How can I get information about HP distributed Smalltalk?

Newsgroups: comp.lang.smalltalk
From: [e-mail] (Karlheinz Blank)
Organization: FAW (Research Institute For Applied Knowledge Processing, Ulm, Germany)
Date: Wed, 25 Aug 1993 07:10:34 GMT

You can get information about HP-Distributed Smalltalk From:

Public Relations Department
3000 Hanover Street
Palo Alto, California 94304
[phone]

Karlheinz Blank, Research Institute for Applied Knowledge Processing
Internet: [e-mail]

---

6.0)     [Smalltalk implementation issues]

---

6.1)         Is Smalltalk interpreted or compiled?

    [Be prepared for more material on this issue. I've got tons of
it, staring at me at this very moment. I'm starting off with this one
mainly because it's short... again, please feel free to submit. In
particular, the whole issue of just what constitutes "interpreting" is
still debated.  --crl]

From: [e-mail]
Newsgroups: comp.lang.smalltalk
Subject: Re: Smalltalk on PC/Compiling
Date: 12 Jun 93 17:55:40 GMT
Organization: ParcPlace Systems, Inc.

    [a poster who requests anonymity] writes:

...
>>4 - Is there a SmallTalk compiler available for Dos?
>    Not as such... Smalltalk is a P-Machine
...
>>6 - Is there such a thing as a SmallTalk compiler?
>>
>    As I said above, it's a P-Machine, not exactly a compiler.
...


The term "P-machine" implies that Smalltalk implementations interpret
pseudo-code. In the first place, interpretation/compilation is not
a language feature but rather an implementation choice. In the second
place, ParcPlace Smalltalk does not interpret bytecode.

What happens as a result of a message send in ParcPlace Smalltalk is:

    - first the system checks to see if the method has
      already been translated to machine code that has
      been cached in memory
    - if the native machine code form is in the cache,
      the system executes that machine code
    - if the cache doesn't contain a translated form of
      the method, the system dynamically compiles the method's
      bytecode into native machine code, storing into the
      cache, and then executes the code from the cache.

Dynamic translation yields the benefits of the execution speed of
compiled code and the space compactness of bytecode. If all the code
in a running Smalltalk image were kept purely in the form of compiled
machine code, the image would consume 5-10 times as much memory, and
therefore could in fact degrade performance on a virtual memory system
by causing increased paging.
--
Michael Khaw    [e-mail] (or [e-mail])
ParcPlace Systems, Sunnyvale, CA    PRODUCT INFO: [e-mail]

---

End of Smalltalk FAQ

--
-----------------------------------------------------------------------------
Craig R. Latta          |  Experimental Computing Facility, UC Berkeley (XCF)
Composer and            |  Atari Games Audio Group         <[e-mail]>
    Computer Scientist  |  The NetJam Project            <NetJam-request@XCF>
[e-mail]  |  The Smallmusic Project    <Smallmusic-request@XCF>
(standard  disclaimer)  |  Biggles' Home for the Oversubscribed (wanna join?)
-----------------------------------------------------------------------------