Slashdot Mirror
Visual Age for Smalltalk For Non-Comm Use
paugq writes " IBM has released Visual Age 6.01 for Smalltalk for free for non-commercial use. You can download it here, it's available for Windows, Linux, Solaris, AIX, etc (registration needed). Smalltalk is a dynamically typed object oriented programming language designed at Xerox PARC during the 1970s"
So before I go through the hassle of signing up for an "account" with IBM, and then downloading and installing the software, could someone please provide a pointer to the information about how it's licensed for non-commercial use?
I've been using Squeak for a number of projects and general areas for 4-5 years now. For me, Squeak is many things- it's my scripting language, my prototyping language, my implementation language (occasionally with a C library when I need the speed), my desktop environment for both my desktop and my PDA. VAST would have a hard time filling all of these needs.
:)
I've used VAST before, and don't get me wrong, it's a really nice product- a rock solid Smalltalk that does a good job fitting in with apps written in other languages the OS supports. I used VAST interning a Progressive Insurance, where it's used for a couple really neat in-house apps. VAST is a really mature development system for web apps, prototyping and final implementation of desktop apps. I've used it in conjunction with the GemStone/S Object Database, which was a lot of fun.
For groups, VAST or VWNC along with ENVY/Developer simply can't be beat- by Squeak or any other development environment/language. It's the slickest way for a group, big to small, to develop an application with relative efficiency. Like CVS, but snazzier.
But for me, it's way too big. I imagine the license is more restrictive as well. It also supports fewer platforms than does Squeak, or even VisualWorks (which also has a non-commercial version for download).
Working toward a usable PDA environment in the spirit of Newton OS: Dynapad
You see, in a Smalltalk environment, the entire system is available to you, at your fingertips.
The same is true of Lisp. I can look at the function slot of the + symbol to get the function. It's a compiled function object. I can use that object opaquely, or if I want to analyze it, I can call function-lambda-expression to recover the source (if it was compiled with debug info), or disassemble to get the assembly code.
The whole system is out there, in the open. I can change it, I can query it.
One company that made their living on Lisp was called Symbolics. They made Lisp Machines, aka lispms. These were specialized hardware platforms with processors designed to run Lisp very efficiently. Also, they came with an OS called Genera. Genera was a programmer's dream OS. I won't try to describe it here, because I couldn't do it justice; I'll just say that I started using Squeak because it reminded me of Genera.
Now, lispms were all about the openness. You could play with the smallest internal function you wanted to. They also came with almost all the source (I think missing just part of the boot loader). But that openness was bordered by your license agreement; you still couldn't share it with your neighbor.
Symbolics wasn't the first to make lispms. They started at the MIT AI lab. This was about the time that parts of the computer industry discovered that Lisp was a good thing. There was a sudden explosion of Lisp in the industry (much like the .com boom). Some of the lab hackers, led by Richard Greenblatt, formed a hacking company called Lisp Machines Incorporated, or LMI. This company was set up to support hackers, rather than maximize profits. (For this reason, they didn't take outside investors.)
While the company wasn't profit-driven, nobody wanted it to piledrive into the ground. So Greenblatt called in Russell Noftsker, who had more business sense. Nofstker promptly took Greenblatt's ideas and started his own company, Symbolics. The two were instantly bitter rivals.
Everybody at the AI lab was associated with one of the two companies, except for RMS. (Symbolics was entirely proprietary, and LMI-- while they wanted an open OS-- was fine with proprietary apps. RMS wouldn't camp with either of them.) LMI wanted hackers to work part-time at the lab, and part-time for LMI. The point of LMI was to support the lab hackers, and the culture. But Symbolics was about making money, and didn't care about the culture. They hired away all the hackers they could. Symbolics also leveled accusations about conflict of interest, and made all the LMI hackers leave the lab.
Almost overnight, the AI lab had been decimated. There's more to this story, but I think you get the idea. The once-open lab culture had been destroyed by ideas of proprietary business.
RMS was very sad to see this happen. He was the only hacker who hadn't left for LMI or Symbolics. This is why he started GNU: he wanted to recreate the culture of openness and freedom that he had known before.
I do believe that the Sapir-Whorf hypothesis applies to programming languages; the design of the languages a programmer knows will strongly influence how he thinks about problems. But this story is a counterexample to the idea that openness in a language design engenders openness in other areas.
I suppose it would come as no surprise that I also consider the JDK to be a colossus, as well. Squeak, compared to other production Smalltalk environments, is a veritable juggernaut, coming with a large array of semi-complete applications built into the default image, including a 3D animation tool, a web browser, email client, irc client, midi player, unused truetype font support, two separate GUI's and heaven knows what else.
I find Squeak interesting, but for serious tasks, its overfeatured standard image makes it a less than ideal choice for me. There are, of course, multiple projects in the works for paring down the size of the image, but, as with many things in Squeak, there has been no clear consensus on how to do it.
Weapons of Mass Analysis
First off, I need to state that I'm a HUGE smalltalk fan/user/zealot....but....
;)
;)
;)
Its a 'monolithic kernel' architecture.
There is of course the "stripping" process to pare down the image size, but frankly it still comes out pretty large
Of course, with the price of memory these days, its not so much of a concern anymore
Seriously though, there was at one point some discussion of an embedded 'cpu', much like the BasicStamp system that would allow smalltalk on 'anything', not sure if it went anywhere
The other problem is one of Education! Very few (and i haven't checked recently) schools offer any educational courses that focus, let along USE Smalltalk.
Text books frequently use banking examples to attempt to show the advantages of using pure objects -- unfortunately banking requires near real-time performance, and traditionally most interpreted systems do not perform. Of course we have bigger badder processors to make up for that
Also, the track-record for large-scale systems hasn't been what i'd call complimentary: Having spent some time working on "!ntegrator" for QWEST (then USWEST), I can tell you that the approach taken was next to dismal in terms of results after three years. Much of which is or has been replaced with either Web-based systems or J2EE.
Objective-C is a great migration/hybrid -- NeXT, and now hopefully Apple will show that.
Now, to its credit, and not to denigrate it any, I kind of see Smalltalk as a "LOGO for Adults" -- seems to be a good sort of tool for instruction.
Its safe to play (sandbox), responsive, and teaches some very important principals (like THINKING before you code
Old age and treachery almost always overcome youth and skill.
I wish they'd made it open source and released it - I think it would be quite a bit nicer even now with a bunch of people poking at it.