6111231@PUCC.BITNET (Peter Wisnovsky) (12/02/86)
I just attended a lecture called "Why object oriented languages are hard
to learn", although it might have been better titled "why smalltalk 80
is hard to learn". It was given by a person from PARC called (I think)
Micheal O'Shea, talking about the results of a study into why so few
people are using Xerox's pet language. Basically, the lecturer had
found that very few people (he said only one that he knew of) had
really mastered the whole (all 2000 object types and 500 message types)
and that this was the reason that only (he claimed) 4 major software
packages had been implemented in the language. He also found that
the problem was not the basic concept of concurrent objects, but that
1: the metaobjects
2: the size
3: the difficulty of getting a realtime graphical view of objects
4: the size
were among the major problems (particularly the size: unfortunately I
did not take notes). He said that a group he is working with is developing
a new implementation to be called deltatalk that will have integrated
metaobjects, be much less extras, and have graphics viewing tools:
oh yes, the syntax (casing, mathematical stuff and general stuff) would
also be simplified and standardized. He also talked about other new languages
that used object oriented stuff like Object Logo, Object Pascal, and interlisp
(I don't know how new that is) and implied that they had greater real
functionality because they had the basic concept but not the difficult
to learn masses of st80 functions, and that he had found that programmers
in such situations will generally not take advantage of these more elegant
constructs anyway because they are too difficult to learn.
As a programmer who has been thinking that objects might come in handy
for implementing certain projects I am working on on a Mac this information
told me:
Don't bother with Mac ST80: one needs 4meg RAM and 300 meg disk space
*Try object pascal (the mpw version)
*Wish for Lightspeed pascal with objects (anyone know when?)
*take a look at object logo for an introduction to object oriented programming.
any comments?
Peter Wisnovsky
Virtual Address: UUCP: ...ihnp4!psuvax1!6111231@pucc.bitnet
Physical Adddress: 179 Prospect Avenue
Princeton, New Jersey 08540
(609)-734-7852jans@stalker.gwd.tek.com (Jan Steinman) (12/04/86)
In article <1414@PUCC.BITNET> 6111231@PUCC.BITNET writes: >I just attended a lecture called "Why object oriented languages are hard >to learn", although it might have been better titled "why smalltalk 80 >is hard to learn"... talking about the results of a study into why so few >people are using Xerox's pet language. Wait until Digitalk has 100,000 copies of Smalltalk/V out there! >Basically, the lecturer had found that very few people (he said only one >that he knew of) had really mastered the whole (all 2000 object types and >500 message types)... (You got that backwards -- there are about 200 classes, and about 5000 methods in ST80.) I don't see what this proves. How many people have "mastered" all the Unix/C library functions, yet useful work (some would claim :-) is still done in C. I think quite the opposite is true of Smalltalk, that it is easier to get useful things implemented without having "mastery" of the whole environment. Besides, I don't want to clutter my mind with all that stuff when it is so easily accessible in a Browser! >...and that this was the reason that only (he claimed) 4 major software >packages had been implemented in the language... How many "major software packages" were available for C within three years of its general availability? How about Pascal? Given languages of similar ages, how many packages are avaliable for Modula-2 or Ada? Besides, just what is a "major software package"? I think your lecturer is badly out of touch -- I know of a dozen or so "major software packages" within Tek alone, and many of our customers have done wonderful things in Smalltalk. If your lecturer means that Xerox only has 4 packages, well, that's their problem! You specifically said "implemented", as opposed to "available". It is true that there is not much "available" software in Smalltalk, but that is because Smalltalk lacks "critical mass", not because of deficiencies in the language. Now that is is available for $99 for the most popular PC things should change. >...He said that a group he is working with is developing a new implementation >to be called deltatalk that will... Aha! The real motive for Smalltalk bashing! If he thinks that a brand new language will fare better in today's market, he better be prepared to give it away so people will use it. Not that people shouldn't continue to develop languages, but any new ones will share Smalltalk's major shortcoming: lack of user base. >Don't bother with Mac ST80: one needs 4meg RAM and 300 meg disk space This is a gross exaggeration. 16 bit Smalltalk functions quite well on the Tek 4404 (68010 based) in 1M RAM with 40M disk. It is true that Smalltalk is not (yet) for everyone. Cheap implementations for popular machines combined with plummetting memory and disk prices will soon cause an explosion in Smalltalk activity. :::::: Artificial Intelligence Machines --- Smalltalk Project :::::: :::::: Jan Steinman Box 1000, MS 60-405 (w)503/685-2956 :::::: :::::: tektronix!tekecs!jans Wilsonville, OR 97070 (h)503/657-7703 ::::::
rfl@oddjob.UChicago.EDU (Bob Loewenstein) (12/08/86)
Try looking a Neon for the acintosh.