home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Language/OS - Multiplatform Resource Library
/
LANGUAGE OS.iso
/
smaltalk
/
irc.lha
/
irc
/
st_irc3_3_92.txt
< prev
next >
Wrap
Text File
|
1993-07-24
|
30KB
|
717 lines
*** this is a transcript of the Smalltalk meeting of March 3, 1992 ***
*** this meeting took place on the Internet Relay Channel (irc) ***
*** I did a bit of editing, you may wish I had done more ***
*** mjb@netcom.com ***
<Ronalda> Go ahead Automan
<automan> We at Cold Spring Harbor Laboratory are working on a rather large
project
*** aknight has joined channel #Smalltalk
<automan> It involves the use of a database from Servio called Gemstone
<mjb> (welcome aknight)
<automan> We will be using Smalltalk to implement a Human genome mapping
application
<aknight> hello
*** tcl has joined channel #Smalltalk
<automan> What I am looking for are implementations of networks. Has aNYBODY
<automan> Wupps.
<Ronalda> Hello Allan. I think I remember Peter and Danny introducing us once.
<automan> Has anybody seen any code on implementing networks (the mathematical
kind?)
<mjb> welcome tcl
<aknight> Yes. I think so. Why are you Ronalda?
*** khaw has joined channel #Smalltalk
<Ronalda> I've seend several. Every one I've seen though would have been
better done using the real objects of the problem domain.
<Ronalda> Hi Mike!!
*** Signoff: khaw (Leaving)
<mjb> (welcome khaw)
<Ronalda> I'm Ronalda because my other choices for nicknames "Carl" and "Opus"
were already taken.
<automan> Well we want to use the networks to represent maps.
*** Ronalda is now known as Carl
*** richard has joined channel #smalltalk
<mjb> welcome richard
<Carl> Hello Rich. Is Adele going to join us?
<richard> Adele will be here soon
*** Gregster has joined channel #Smalltalk
<Gregster> Hiya!
<mjb> hi Gregster
<Gregster> Hey MJB!
<Carl> Back to the Networks problem. Why not implement the objects in your
real domain. "Maps" or whatever you called them?
<mjb> *** quick 1 ? survey: what ST is everyone using ****
<richard> ST V/
*** rayn has joined channel #Smalltalk
*** Signoff: tcl (eff.org)
<mjb> ST V/Win
<Carl> Smalltalk-80
<automan> The maps vary in the kinds of data they represent, we want a generic
means of representing them
<automan> Smalltalk-80
<mjb> hi rayn
<aknight> We're using ST-80.
*** tcl has joined channel #Smalltalk
*** Signoff: tcl (tcl)
<rayn> Hello - this is my 1st IRC. I'm using ST V/Win
<mjb> welcome rayn
<Gregster> HUh?
<Gregster> What's with this smalltalk thing?
<automan> We are also just looking for examples of code that do networks too.
We have only been using Smalltalk for a few months and so we're always
looking for good examples.
<Gregster> <----VERy VERY confused!
<automan> Gregster, I was asking if anybody had examples of mathematical
networks.
<mjb> Gregster - ST is an object oriented programming language/environment
<Gregster> LIke?
<Carl> Gregster: Do you know what the computer language Smalltalk is then you
probably don't want to be here.
<Gregster> See ya!
<mjb> adidas
*** Gregster has left channel #Smalltalk
<richard> is there a limit of 10 people to a conference, as the help messages
state?
<Carl> I'm going to change the topic string to something more descriptive.
*** Carl has changed the topic on channel #Smalltalk to Chatting about
Smalltalk, the programming language
<mjb> seems like I've seen crowds larger than ten on these channels
<aknight> Automan: Have you looked in the ftp archives at uiuc.
<Carl> I don't think there is a limit on simultaneous users.
*** peaches has joined channel #Smalltalk
<automan> aknight: Yeah, and I found a really good file in on Directed Graphs,
but nothing on networks.
<mjb> hi Peaches
*** Signoff: rayn (Ping timeout)
*** tcl has joined channel #Smalltalk
<automan> Has anybody gotten on ParcBench and seen anything on networks?
<Carl> I'd like to chat about what people think the future direction of
Smalltalk should be.
*** rayn has joined channel #Smalltalk
<richard> Hi--Richard Dellinger is VP Engineering ParcPlace. We are actually
two people here together and I am Adele Goldberg. ParcBench does not have
something on networks at this time.
<richard> And specifically I would be interested in learning what Smalltalk
programmers would like in the way of Team Tools.
<Carl> Alan, doesn't Peter and Danny have implementations of general networks?
<automan> richard: code sharing
<aknight> Carl: It's possible. I'll ask them. Automan: perhaps you could
leave me your e-mail address.
*** peaches has left channel #Smalltalk
*** jsp has joined channel #Smalltalk
<mjb> Welcome Adele
<aknight> richard: We've just gotten ENVY for Smalltalk-80 and I'm very
impressed. Pricey if you're not a University, but then so is ST-80.
<mjb> hi jsp
<tcl> How about debating native gui vs. cross-platform-consistent gui in st
*** jsp has left channel #Smalltalk
<richard> Of course, if you join the Partners Program the price is quite
inexpensive. The idea for the Partners Program is to provide easier access
as well as future assistance in marketing any products you might create.
<Carl> I think maintaining a native, Smalltalk implemented, GUI is desirable
for Smalltalk
<automan> tcl: I like the portability that comes with
cross-platform-consistency, but I know the community that I am programming
for and they want a program running on a Mac to look like a Mac program, and
the same goes for the other platforms.
<aknight> Carl: What do you mean by native, Smalltalk implemented. Do you
mean that Smalltalk re-implements e.g. Motif or that Smalltalk is its own
thing (like pre 2.5) or just that ST has an abstracted view of the native
GUI.
<mjb> richard, would you define quite inexpensive?
<richard> This is not an either/or issue. Platform compliance is important,
especially for people who have fairly homogeneous computing situations. In
addition, some way to augment the platform specifics where you have neat new
ideas is also key.
<Carl> I mean that Smalltalk does its own thing (like pre 2.5). Its
unfortunate I think that we don't have a native Smalltalk pratform.
<richard> Inexpensive is the same price as educational $350. However we also
ask members of the Partners Program to get support. This brings the price to
about $800 (comparable to Borland's C++ without support).
<aknight> I didn't like the 2.5 approach. I don't like having to choose
between using Smalltalk or any of my other tools.
<mjb> thax
<mjb> er... thanx
<Carl> Yes, on a Smalltalk platform that issue would go away. All the tools
would be in Smalltalk.
<aknight> Carl: like the Momenta.
<aknight> Carl: The choice is still there, it's just do I walk over to the
dedicated Smalltalk machine and forsake my tools, rather than do I open up a
Smalltalk window and forsake my tools.
<Carl> True. All my tools were in Smalltalk. And I also enjoyed being to
able to experiment with new UI concepts that restricting myself to the
abilities of the OS's GUI often restricts what I can do.
<richard> Let me try again on Team Tools. Do any of you work in groups of 4
or more Smalltalk programmers?
<mjb> not I
<automan> richard: Currently we have three full-time programmers plus 2
mathematicians
<mjb> rayn, how are you doing with that st v?
<aknight> Technically, I work with 4 or more, but most of them are only
part-time Smalltalk and many are students. There's not a lot of
coordination, something we're trying to change.
*** hanz has joined channel #Smalltalk
<mjb> welcome hanz
<hanz> hi
<mjb> what St are you using
<mjb> ST?
*** hanz has left channel #Smalltalk
<rayn> I work by myself, but manage to mis-coordinate anyway by trying to move
more than one project ahead at a time. I'm a relative newcomer to Smalltalk,
so probably haven't hit the problems of really large projects.
<tcl> machines with st operating systems, and all software written in
smalltalk are a nice dream, but for now we should talk about the reality - we
have to deal with multiple platforms. In our environment i would prefer that
st work the same way on all platforms
<tcl> others would prefer that st work in the native way. Can parcplace
perhaps cater to BOTH needs?
*** gary has joined channel #smalltalk
<mjb> welcome gary
<gary> hi
<tcl> richard: yes, we have four programmers in our teamstem do
<gary> What is being discussed?
<mjb> what ST system do you use?
<richard> tcl: yes, in fact that is the dream. Now let's discuss reality.
We want to maintain full portability and allow you to switch between host
look and feel or your own.
<gary> st80 rel 4.0t
<mjb> gary, then you'll be right at home.
*** Signoff: tcl (irc.mit.edu)
*** Calvino has joined channel #Smalltalk
<richard> Switching can not mean emulation...chasing changes to the host bits
is simply impossible. So we have to devise techniques and technologies for
setting look and feel policies and then calling on the host bits where the
policy dictates.
<mjb> hi Calvino...
<Calvino> Hi all
<mjb> you made it
<Calvino> Yeah!
<aknight> richard: That sounds really interesting. How far along the road is
this? 4.1, 4.2...
<richard> Not 4.1, which is in beta right now.
<aknight> Any chance of a preview of differences 4.0->4.1
<Calvino> mjb: During a band rehearsal, no less.
*** HaiLuv has joined channel #smalltalk
*** tcl has joined channel #Smalltalk
<HaiLuv> HI everyone....
<mjb> welcome HaiLuv
<tcl> why do you folks kkeep sighing off and on?
<mjb> HaiLuv, what St system do you use?
<HaiLuv> any interesting conversation goin on here....
<mjb> Smalltalk programming language is the subject here
*** Calvino is now known as Craig
<richard> Sure. 4.0 was a big change, so there was a bug tail to chase. Lots
more documentation, especially on graphics. Additions to the graphics
package. A special capability to call on and call from C with automatic
class creation and dynamic linking.
*** dpr has joined channel #Smalltalk
<HaiLuv> Well I guess I'm outa here.....don't know what that is...cya....
*** HaiLuv has left channel #Smalltalk
*** kmorriso has joined channel #Smalltalk
<mjb> welcome to the Smalltalk programming language channel
<aknight> Richard: Sure, document it now, after I've gone and figured it out
the hard way.
<richard> Continued on 4.0: Dynamic linking is of course only on platforms
that support it. Also automatic type conversion of parameters and return
values. Some additional mechanisms to facilitate creating pluggable parts
(we call these value holders).
<mjb> dpr, what St system do you use?
<mjb> ST?
*** dpr has left channel #Smalltalk
<mjb> kmorriso?
*** Bejeezel has joined channel #Smalltalk
*** Signoff: kmorriso (kmorriso)
*** Bejeezel has left channel #Smalltalk
<aknight> Hmm. I like the ability to call from C. Opens up a lot of
possibilities.
<Craig> Back in 5
<richard> Alan: what would you do with the call backs?
<tcl> msg tcl test to myself
<mjb> try "/msg whoever"
*** MikeOB has joined channel #Smalltalk
*tcl* yup, i mistyped
<mjb> welcome MikeOB
<aknight> We do finite element analysis. Big codes, much of it in FORTRAN.
The ability to call Smalltalk modules to do things would be very nice. In
particular, our remeshing code is written in Smalltalk. Currently, that's a
prototype, but if we could actually
<aknight> call it directly that would be very nice.
<richard> Neat. If you can call from your Fortran to C, then you will be able
to call from Fortran to Smalltalk Rel 4.1 with this external programming
option.
<aknight> Great. That may yet save us from C++.
<gary> good night all. I'm signing off.
*** Signoff: gary (Leaving)nite ga
<tcl> Aha, so you talk about rel 4.1 --- let's hear some more details of
what's in it!
<MikeOB> I seem to have missed something by coming in late - Rel. 4.1 is
actually going to let us do callins?
<tcl> is 4.1 going to be easily shrink-wrappable?
<richard> tcl: we did this...is there a log to review?
<mjb> I'm loggin
<mjb> I will post log to (?)... suggestions
<aknight> One feature I'd like in ST is a profiler that would give me exact
numbers for each statement (or at least method). Probabilistic profilers are
nice, but there's a lot of good uses for line-by line ones. I know it would
take forever to run, but I'd
<aknight> still like it. When I get some (hah!) spare time I may yet write
it.
*** peaches has joined channel #Smalltalk
<mjb> hi peaches, welcome, to the Smalltalk programming ;anguage channel
<mjb> language
*** LunarWolf has joined channel #Smalltalk
<LunarWolf> merry meet
<MikeOB> I suggest that since we all must have TCP capability (or we wouldn't
be here) that it makes sense to make the log available for anonymous FTP.
That'd save on Usenet bandwidth.
<Craig> I can offer a site.
<Carl> I'm logging the conversation right now.
*** LunarWolf has left channel #Smalltalk
<tcl> On the topic of profilers, I'd like to se two features: breakpointing
(without editing to add a self halt) and inst var update/access tracing or
breakpointing
*** WintrMute has joined channel #Smalltalk
<mjb> my understanding is that there is a ST archive at speedy.cs.uiuc.edu,
does this ring a bell with anyone?
<aknight> I'd like to second tcl's requests.
<aknight> mjb: Yes, there is an archive. Aliased to st.cs.uiuc.edu
<WintrMute> whats up people?
*** Signoff: peaches (poe.acc.Virginia.EDU)
<automan> I'll third that, I have used Servios breakpoint browser and find it
really useful
<MikeOB> Hmmm. The ParcPlace debugger has the ability to execute up to the
carat in the current method, but it's not clear that the architecture of the
VM would allow variable access trapping - sure would be nice though.
<richard> good ideas, are there any other requests to improve the debugger
*** Signoff: WintrMute (Leaving)
<mjb> WinterMute, this channel is for the discussion of the Smalltalk
programming
<MikeOB> One poor second I put in was a "debug" message in Object, which did
nothing unless left-shift was down, then it did a halt. A rather nice
conditional halt.
<aknight> richard: In Smalltalk/V, when you set a breakpoint in a method, it
stops whenever you enter a block defined in that method. That was quite
handy for quickly getting into the guts of e.g. detect:, and I miss it.
<richard> Alan: are you talking about having a breakpoint that triggers
whenever a block in a method is evaluated? ...
<mjb> a few questions...
<mjb> How often would like to have this meeting? weekly? bi-weekly? monthly?
never again? :)
<mjb> What day?> What day?
<mjb> Time of day?
<mjb> Time of day?
<aknight> richard: Yes. e.g. whenever a sortblock or the block argument to
an iterator is entered. In /V this always happens when you set a method
breakpoint, so it's probably a semi-accidental "feature".
*Carl* I think monthly would be good.
<Craig> mjb: I like late evenings (here). Then it's a decent time overseas.
<tcl> here in eastern time it is 22 past midnight. Future irc sessions
should be earlier. We are losing a lot of europeans too.e is here?
<mjb> where is here?
<aknight> mjb: For time of day I suggest morning Pacific time, which would
allow people from Europe to participate.
<automan> mjb: Bi-weekly would be nice, and so would an earlier time for
those of us on the east coast.
<richard> The blocks in V are not full closures as they are in Smalltalk-80.
<aknight> richard: Yes. To do it right you should probably be able to set
breakpoints on entry to a particular closure.
<automan> mjb: Could you post to News the name of the site where the log of
this discussion will be?
<mjb> sure, but what's the definition of News#Smalltalk
<automan> mjb: Usenet Newsfile location to comp.lang.smalltang.smalltalk
<mjb> I was plan to post the file location to comp.lang.smalltalk
<richard> We have to go now. See you again.
*** Signoff: richard (Leaving)
<MikeOB> I'd like to point out that while variable tracing is a wonderful
idea - I think it's the single greatest improvement the debugger could see -
that code is global but instance vars are local. You'd have to specify
which object as well as which inst var was for coming
<tcl> presumably you mean comp.lang.smalltalk
<MikeOB> to be traced.!
<mjb> thanx for coming!
*** peaches has joined channel #Smalltalk
<automan> mjb: Thanks. I have to go now too (meeting in the morning)
<aknight> It would also be useful to have a breakpoint when a variable is
modified in any instance of a particular class.
<rayn> Do any of you veterans use any automated tools for visualizing
complete Smalltalk "applications"?
<Carl> Only Projects
<tcl> Could some of you parcplace folks post more details about 4.1 to
comp.lang.smalltalk, some of us have no access to ParcBench
*** automan has left channel #Smalltalk
<Carl> I'll write it down as something to do.
<MikeOB> aknight: I agree, but most variable tracing is done by putting
breakpoints on a particular address (at the machine level). This would still
work in ST80 but would limit such tracing to particular objects. To put on a
general trace on a particular inst
<MikeOB> var in all objects would require chasing through all of object memory
setting hardware breakpoints.
<aknight> MikeOB: Well, it's pretty easy to find all methods which might set a
variable, but it could be a fair number of breakpoints.
<MikeOB> Yes, there's nothing against it, really, but all the chasing and
bookkeeping. It would be like a mark-and-sweep GC in terms of overhead.
<Craig> Arg. Can't do this and rehearse at the same time... I'll be watching
comp.lang.smalltalk for the next meeting. later...
*** Signoff: Craig (Craig)
<tcl> You could do var access breakpointing by modifying (temporarily) the
*methods* that already update the instvar (although this would miss
instVarAt:put:
<aknight> I'm willing to make anybody who uses instVarAt:put: suffer.
<tcl> well i'm gone, its too late in the east ...
*** Signoff: tcl (tcl)
*** Huyck has joined channel #Smalltalk
*** Signoff: Huyck (Bad link?)
<mjb> well...
*** Signoff: MikeOB (ucsu.colorado.EDU)
*** Signoff: Carl (ucsu.colorado.EDU)
<aknight> I've got to get up in the morning. So long all.
*** Carl has joined channel #Smalltalk
*** MikeOB has joined channel #Smalltalkback
<mjb> back?
*** aknight has left channel #Smalltalk
<mjb> guess it's time to put this puppy to bed, eh?
<MikeOB> I guess so.
<rayn> Are there any thoughts about what the "next" smalltalk might look like?
<Carl> Gee. I thought we'd have more to talk about.
*** Rohan has joined channel #Smalltalk
<Carl> The "Next" Smalltalk. You mean the next generation of Smalltalk?
<MikeOB> Maybe it'd go better next time if an agenda were made up and
published in comp.lang.smalltalk.
<rayn> I'm thinking not the next version, but yes, the successor.
*** Rohan has left channel #Smalltalk
<mjb> if we can getmore participants things might be better, still not bad for a
first meeting...
<MikeOB> Well, Self is one entry in that sweepstakes. Some folks seem to like
the idea of templates over classes, as somehow more "unifying".
<mjb> hard to set an agenda when you don't know who will show up...#Smalltalk
<Carl> I think the next generation of Smalltalk will run natively on the
machine. As the operating system itself. This has already been done of a
few machines.
*** eggshell has joined channel #Smalltalk
<eggshell> Hi, peaches!
<mjb> I will post this log to the cis -digital;k forum as well...
<Carl> Yes, SELF is a contender. I personally found Classes easier to deal
with.
<MikeOB> mjb: Well, true, but it would give a direction to the discussion.
<eggshell> Hi, all.
<mjb> it might encourage some DT people to show up as well
<rayn> Maybe an improvement, but Self seems incremental to me. How about
something that would presevre the Idea of Context in a GUI world.hi eggshell,
*** john has joined channel #Smalltalk
<john> Hey all
<mjb> hi eggshell, welcome to the Smalltalk programming language channel
<mjb> welcome john
<mjb> welcome john
*** carrier has joined channel #Smalltalk
<john> thanks mjb
<eggshell> Hi, mjb. Thanks.
<mjb> eggshell, john, what ST environments do you use?
<MikeOB> I'd tend to believe that a future contender would have tools which
would allow some level of programming interaction above the class level but
below the full application level. A few more layers in the hierarchy to make
things more manageable.
<Carl> I'd been thinking of something more GUI oriented. A Smalltalk where
the screen was the root for garbage collection. Its pretty hazy though. I
think just better and more graphical programming tools would take you a long
way there.
*** Alpha has joined channel #Smalltalk
<MikeOB> Sort of the way TCL and Perl have gotten very popular in making UNIX
apps more manageable.
<Alpha> 'lo all
<mjb> welcome to the Smalltalk programming language channel
<rayn> I'd like to see something in a graphical environment that could help me
out the way shell history does.
<Alpha> YIKES-to informed for illiterate 'ol me.
<carrier> greetings all
<carrier> leave #notalk
*** carrier has left channel #Smalltalk
*** john has left channel #Smalltalk
*** Alpha has left channel #Smalltalk
<MikeOB> You know what I like? Bots. Independent assistants which could take
high-level commands and spray out low-level messages to all and sundry. I'm
very impressed with bots.
<Carl> Yes, application-level programming tools would be nice too.
*** Voyeur has joined channel #Smalltalk
<Carl> Are you talking about artifically intelligent "bots"?
<mjb> Voyeur, welcome to the Smalltalk programming language channel
<rayn> How can the environment be set up so I can build a Really Stupid,
simple Bot, like shell history.
<Voyeur> hi and thanks!
<MikeOB> In a way. My fiancee has had her life taken over by MUDs. And there
are folks who have written independent programs which operate in the MUD as
"homunculi" - artificial players. They use the same interface as people do,
which in the
<Carl> What would this Bot do? There's no command line interface to Smalltalk
(thank god).
<MikeOB> case of Muds is not hard since they're text-only.
*** Voyeur has left channel #Smalltalk
<MikeOB> Well, remember that recorders and replayers of interactions have been
used to launch Smalltalk demonstrations, so event strings can be managed.
*** eggshell has left channel #Smalltalk
*** peaches has left channel #Smalltalk
<MikeOB> A lot of work has been done on "user agents" in the past, but I've
never seen anything as capable as a MUD "bot". I'm totally impressed by
them. They act as walking mailboxes, tourguides, etc. If we could come up
with a "bot" that acted in the
<Carl> Yes, I've seen them and helped write them. But I can't see them being
terribly useful for anything other than demos and regression testing. I
think you just want better tools that make doing the common things easy.
<MikeOB> capacity of, say, DWIM and Masterscope in Interlisp, we might really
have something.
<rayn> The problem with recorders and such is that the idea of their "context"
is not strong.
<Carl> What is "MUD"?
<MikeOB> MUD = Multi-User Dungeon. Sort of like IRC with manipulatable
objects - "look" and "take" commands.
<rayn> I think that the whole idea of the "click click click" sort of
interface has to be expanded if what I want is to be built reasonably.
<Carl> Hmmm.... I'd like to see this MUD. The capacity for building these
"bots" you speak of sounds interesting. Where does "MUD"Live?
<MikeOB> Well, consider context evaluators in chess programs. One can
consider that the VM, instead of just looping uselessly for input, or hanging
on a semaphore, could make use of "dead" time in building suggested
alternative strategies.
<MikeOB> The spell checker is a really simple sort of assistant, for example.
<Carl> Yes, I've implemented tools that ran in the background waiting for
useless time and doing things like analyzing the classes and methods in your
system finding bugs and things to warn you about.
<MikeOB> For information on MUDs, send email to my fiancee,
"valerie@ucsd.edu". I don't know enough to be able to point you in the right
direction. Tell her I sent you for info. MUDs are distributed, like IRC,
but the server lives on a single machine which is
<MikeOB> "home" to that particular MUD.
<mjb> how do they determine "useless time"? they have a low priority?
<rayn> MikeOB, alternative stratagies might be a plus, and background helpers
also. But I'm really looking for a canonical discription of the system (UI)
state, that can be manipulated strongly.
<rayn> How can I build a ".cshrc" that would let me look at *your* environment
<MikeOB> Well, but consider that the UI is merely the control panel to the
tools underneath. If the UI is to be changed to provide new services, what
are thos eservices to be?
<rayn> in a way that makes most sense to *me*
<Carl> Yes, using a low priority.
<mjb> thannx
<Carl> MikeOB: I don't get what you mean. The UI IS the tools.
<MikeOB> Another possibility, which we are working on here, is the notion of
"wrapping", wherein pieces of a system are described in a database which
enables them to be assembled automatically into larger tools.
<MikeOB> Carl: I disagree. A UI is a window to underlying capabilities, and
is most functional when least noticed. That's my philosophy, anyway.
<rayn> This description sounds closer to what I'm looking for. Could you say
a little more about it?
<Carl> My philosophy is direct manipulation UI. Where from the users point of
view, the UI IS the tool. At least that is the metaphor that the programmer
sweats to try and achieve.
<MikeOB> We're still working it out. Basically, one application is for a
planner to make use of a planning database to formulate a path to a goal,
using a resource database, then use the wrapping database to assemble the
resources into a single entity which can<MikeOB> achieve the goal.
<rayn> Carl, I agree with you for most cases, but sometimes I wish to describe
a UI and how to use it to some robot.
<Carl> MikeOB: This wrapping database sounds conceptually deep. You would
probably have to describe a lot more about it before I could get my mind
around how it works.
<MikeOB> So, Carl, you are looking for better ways of achieving the metaphor
of direct manipulation, and I am worrying about what the tools could be.
Complementary, I think.
<MikeOB> It's very deep and our brains are frying, but we're building simple
examples using FORTRAN models, RPC, message busses and a lot of what we call
"C--" to glue it together.
<Carl> God, I never though about have a way to describe to a software robot
how to manipulate the interface.
<Carl> Good heavens! With that mismash of languages be careful you don't
create a software frankenstein!
<MikeOB> Oh, I forgot the Smalltalk, the Prolog and the C++.
<MikeOB> Actually, RPC saves the day. We switched the UI from a Smalltalk
prototype to a C++/Interviews package without changing a line of code
elsewhere in the system. RPC saves!
<rayn> It's my prediction that just like spreadsheets made everyone
programmers, and desktop publishing made everyone typesetters, that we are
coming on to an age of desktop systems integraion. Similar advantages and
problems.
<Carl> Has anyone here seen Nicklaus Wirth's new language, Oberon?
<MikeOB> It was a lot of fun doing RPC from Smalltalk before Smalltalk could
really talk to the net... But that's sort of off the topic.
<MikeOB> No, tell us about Oberon.
<rayn> RPC is one part of the kind of thing I'm thinking about.
<Carl> It's a pure garbage collected language like Smalltalk. It has the most
bizarre but powerful UI mechanisms I've ever seen. It implements ALL the UI
itself (like earlier Smalltalk's did). It was quite fascinating. There are
implementations of it for
<Carl> several machines. I was using the Mac implementation. I didn't get to
play with it for a long time but what I saw was fascinating. A fully
manipulable UI with integrated support for multimedia.
<rayn> Thanks. I'll look at Oberon.
<MikeOB> Where can one find more info on Oberon? Is it a product or a project
or a prototype?
<Carl> I think its a project by the University that Niklaus is part of. I got
the info on it about 8 months ago from the net. I can't remember where it
was. I wish I could find it again so I could explore it some more.
<rayn> I have seen freely distributed version for the IBM PC.where get
<rayn> !archie -s oberon
<mjb> where get that be obtained?
<mjb> ah
<Carl> Yes, the versions I saw for Suns, and Macs were freely distributed too.
*** Signoff: Carl (Bad link?)
<mjb> guess he got dumped
<MikeOB> Wups. Bye carl!
*** DeadBoy has joined channel #Smalltalk
<DeadBoy> HELLO????
<mjb> DeadBoy, welcome to the Smalltalk programming language channel
<MikeOB> Hmmm. If Carl doesn't make it back soon maybe this is a wrap.
<mjb> yep
<MikeOB> I love it. Bots in Smalltalk. What a concept.
<mjb> I'd like to explore that more
<DeadBoy> SORRY WRONG CHANNEL. OOPS
<DeadBoy> BYE
<mjb> software 'bots, that is
<MikeOB> I always did think little assistants running around sweeping up
object space and putting things in boxes would be awfully handy.
<MikeOB> One of my big pet peeves is that there are NO integrity-checking
mechanisms around any more for images.
*** DeadBoy has left channel #Smalltalk
<mjb> I have a lot of image problems in st v/win
<rayn> Besides the Bots themselves, I'm hoping for something like syntax or
union rules so a real Bot bueracracy can develop and be controlled.
<mjb> the st v/pm is supposed to be a much better product
<MikeOB> It used to be that doing an image "clone" (writing a new image file
using Smalltalk code to clean things up) would do nice things, but no more.
When Xerox found all that deadwood hanging around under the Dependency class
variable in Object I knew there
<MikeOB> was trouble in river city.
<rayn> Where do you follow St v/win developments? CIS?
<MikeOB> Enough ranting. I should go to bed. Catch you next meeting.
Looking forward to the log of what I had to miss! G'night all.
<mjb> CIS mostly, BIX also
<mjb> night
*** MikeOB has left channel #Smalltalk
<rayn> Thanks for the chat, folks. I'll watch comp.lang.smalltalk for the
next announcement.
<rayn> bye
later
<mjb> later
*** Signoff: rayn (Leaving)
/quit
$