SciFi history?

[DMcCunney]

Originally Posted by mike_bike_kite

The idea of RPN is that you don't need parenthesis however LISP is famous for requiring thousands of the damn things - what gives?

Let's draw a distinction between the syntax requirements of the language and those of things written in it.

I suspect AutoLISP uses RPN because it's easier to parse when evaluating.

Modern languages like Java are highly optimised and I find I can write board games like Checkers, Connect 4 or Othello and performance is never a problem - hell you can even write arcade programs in it. No doubt Fortran might be a little quicker but then I also have to factor in the time it takes me to write the programs. I'm not even sure if you can get assemblers these days for computers - I assume that C is the lowest language you could use.

Oh, you can get assemblers - the Gnu Compiler Suite includes the as assembler, for example. The question is why you might need to use them. Hardware is fast and cheap enough that it it's no longer necessary to write in assembler to get performance, and about the only reason I can see for resorting to it for writing device drivers or for embedded applications.

I thought whitespace was very clever. I've written a few small languages myself and just wish I had that idea

I wouldn't actually try to code in whitespace, but I was tickled by the idea.
______
Dennis

[DMcCunney]

Originally Posted by kacir

When you write a program in AutoLisp and you want to prevent other people from reading the program you simply strip all the CR/LF characters (make it one long line), replace tabs (used for formatting) with spaces and then replace every occurrence of more than one consecutive space with one space. When you replace all the variable names with some A1, A1, A3, B1 meaningless names, it is guaranteed that nobody will be able to understand your program without LOTS of effort, especially when the program is long ;-)

An old friend programs in APL. At one job, they had to send their code to someone else to run. Since APL was interpreted, they had to send source, and there were issues about intellectual property. His solution was to simply strip the comments. The lawyers looked at the result and pronounced it object code, and okay to circulate outside the company.

By the way, do you know the GNU Emacs text editor?

Eighty Megabytes And Constantly Swapping?
Yes, and an assortment of things like it. In normal usage, I'm more likely to use vi.
______
Dennis

[theducks]

Originally Posted by nyrath

There are circular slide rules for sale here
http://www.sphere.bc.ca/test/circular.html
However, I'm trying to make an astrolabe using a laser printer
http://www.flickr.com/photos/nyrath/...7624308442574/
I've written a python program that draws the scales for me.

I still have a 4" 2-sided Circular rule around here (somewhere) that was easier to use (and carry) than the 12" they issued in Navy ET school.

[mike_bike_kite]

Originally Posted by DMcCunney

Let's draw a distinction between the syntax requirements of the language and those of things written in it.

I suspect AutoLISP uses RPN because it's easier to parse when evaluating.

RPN is definitely easier to parse than standard notation but I don't think that's the reason for the layout. The whole language seems to be built around a simple list parser. If you're going to process the items in a list then you should say what you want to do with each item at the start. The advantage of building the whole language around a simple construct (like a list) is that it makes the language easier to learn and also straightforward to expand. In lisp at least, I don't think there is a distinction between the syntax, the programs and the data.

[DMcCunney]

Originally Posted by Worldwalker

Emacs has everything. It's just that you have to sit on a mountaintop wearing nothing but a loincloth and eating lychees for a year before you can actually use it all.

Not quite. But you pretty much have to customize it to be really productive, and there's a substantial learning curve required to be able to customize it.

OTOH, you can do pretty much anything from within Emacs, including email. Usenet news, and using a command line. I knew folks back when who would invoke emacs as the first action after logon, and just stay in it and do everything from it.
_____
Dennis

[DMcCunney]

Originally Posted by mike_bike_kite

Emacs is a bit flash for me - I've yet to move on from VI.
PS thanks for the Lisp intro.

I have Gnu Emacs (and Xemacs, and an assortment of other things that use emacs as the design concept) here. But what I learned back when was vi, and it's what I still tend to use. The underlying program might be Vim, but for what I do, it might as well be vi, as I seldom touch the features Vim has that vi doesn't.

(I'm a sysadmin, not a developer. I use an editor to diddle config files and write scripts. Emacs would be overkill.)
______
Dennis

[DMcCunney]

Originally Posted by mike_bike_kite

RPN is definitely easier to parse than standard notation but I don't think that's the reason for the layout. The whole language seems to be built around a simple list parser. If you're going to process the items in a list then you should say what you want to do with each item at the start. The advantage of building the whole language around a simple construct (like a list) is that it makes the language easier to learn and also straightforward to expand. In lisp at least, I don't think there is a distinction between the syntax, the programs and the data.

IIRC, AutoLISP is actually based on David Betz's XLisp, which is a superset of the Scheme dialect of LISP.

But yes, LISP is intended to be a list processor. Emacs was originally implemented as a set of macros in the TECO language that ran under the ITS OS on a DEC-10 at MIT. Richard Stallman unified and extended several extant TECO macro packages in common use, and called the result Editing MACroS. They reached the point where everyone used it and often weren't even aware TECO was under the hood.

When TECO was going away, Stallman rewrote in LISP. The nature of the language was a good for the sort of string processing tasks an editor would do, and it was available on the machines where emacs would be used. (Other available high level languages presented implementation issues, mostly because the machines in question didn't have facilities for handling character I/O.)

James Gosling (who created Java) did the first version of emacs in C, implementing a language called mocklisp. Stallman later did his own C language port with a genuine LISP interpreter as the core and most of emacs written in LISP.

One valid criticism of LISP in emacs is that the macro language used in an editor ought to be usable elsewhere as well, and even if you are a die-hard emacs user, what else do you do with LISP? Likely, nothing. Eric S. Raymond (who is a Gnu Emacs contributor) was going on a while back about what emacs got wrong, and thought that Python would be a better underlying language. (There are an assortment of editors written in Python - see http://texteditors.org/cgi-bin/wiki....onEditorFamily for a partial list.) I suggested he should rewrite emacs to use Python and was told "Don't tempt me!"
______
Dennis

[DMcCunney]

Originally Posted by calvin-c

Strange-the terms *I* remember used to belittle the stories I enjoyed 50 years ago were 'trash', 'escapist', and 'science fiction'. I don't recall hearing SciFi used at all until the 80's, when science fiction started becoming popular. (That's when all us long-time readers started turning into pretentious snobs because we'd recognized the worth of thinking about the future years before.)

Whether you heard the term SciFi being used depended upon who you talked to.

The term was actually coined by the late Forrest J. Ackerman, a long time SF fan, editor, and agent, as a contraction of Scientifiction, the term Hugo Gernsbach used to describe what he published in Amazing and Thrilling Wonder Stories. SF fans had profoundly mixed feelings and some never quite forgave Forry for it, because the term as used by folks who weren't SF fans connoted all the things they thought made SF at best second rate literature.

These days, SciFi is the common term, and I'm philosophical, as it's no longer seen as a denigration. It's become popular in the broader market, and there are films and TV as well as books that fall under the genre.

I just tell folks "I watch SciFi. I read SF."
______
Dennis

[Greg Anos]

Originally Posted by mike_bike_kite

...
Modern languages like Java are highly optimised and I find I can write board games like Checkers, Connect 4 or Othello and performance is never a problem - hell you can even write arcade programs in it. No doubt Fortran might be a little quicker but then I also have to factor in the time it takes me to write the programs. I'm not even sure if you can get assemblers these days for computers - I assume that C is the lowest language you could use.

...

The hardware you're running your Java on is probably more powerful than the top line Cray supercomputer in 1980... I hope you can run a checkers programs without bogging it!

I am a dinosaur programmer (COBOL). I took a Java course about 2 years ago, thinking about retreading. Here is the performance differences, both from Object Oriented, (all O O languages) and Java, per se.

O O - How do you change an object? Well... the object has the code that made it embedded as part of the object. You end up re-executing the code to get at the data, change the data, and re-encapsulate the data with the code. You think that isn't overhead? In a non O O language, you read the data raw, in your own data layout (or existing file layout you reuse), mod the data, and write. An order of magnitude less machine use. (But you have to know your data! (i.e. you have to know what you are doing.))

Java - an Interpreted language. Any interpretive language is an order of magnitude slower that a compiled language. (Sometimes more that one order of magnitude). A compiler writer is a special breed (I know just one.), who makes the compiled code efficient. Very efficient.

Garbage collection...Jeez, I thought that went out with BASIC in the early 1980's! IBM solved that problem back in 1975!!! (Had to! How else could you run 5,000 interactive terminals simultaneously on a machine the power of a Intel 486. One machine, 'cause it cost over a million dollars then!)

If you want to get rich as a programmer, figure out how to write a Java compiler that writes out non O O efficient code. Sell to the cloud world, where efficiency save the company money, unlike today's paradigm, where efficiency means nothing, because somebody else is paying for the hardware...

(Falls over after the tranquilizer dart hits, and the men in the white coats come over with the straightjacket..)

[mike_bike_kite]

Most Java "interpreters" are actually just in time compilers which spot where the main loops are and then compile it. If a program runs for a short while then it will all get compiled. End result is a program that runs almost as quick as a compiled program. My board game programs will search 1/2 million board positions in under a second and that's using an antique of a computer. This compares well with assembly language programs that I've written (on much older processors). The garbage collection is quite good in Java - I don't give it a second thought but then it depends on how you write the programs as to how much garbage needs to be collected. Similarly there is no need to use the OO features of the language - I don't

[Worldwalker]

TECO ... Text Editor & COrrector ... the editor I loved to hate, back in the day. What can I say, I liked EDT. Still do. My fingers could probably find the EDT hotkeys on a VT-100 even today, and it's been decades. I used TECO and hated every minute of it. Though the fact that there was an Eliza game written in it was pretty awesome.

The only language I've ever written was a little scripting language called TBL, for a gamemaster's aid program I wrote for D&D players. I didn't plan it out all that much in the beginning, so it kind of just grew. Some of the syntax was taken from COBOL (in its defense, that hideous language was easy for non-programmers to understand), some was from BASIC (ZX BASIC to be exact!), and some from the twisted depths of my imagination, sense of humor, or coding needs of the moment. It worked surprisingly well, all things considered. But it was definitely a bit weird.

RSE, I'm surprised at you! It's straitjacket. It's tight, not linear!

[DMcCunney]

Originally Posted by Ralph Sir Edward

I am a dinosaur programmer (COBOL). I took a Java course about 2 years ago, thinking about retreading. Here is the performance differences, both from Object Oriented, (all O O languages) and Java, per se.

O O - How do you change an object? Well... the object has the code that made it embedded as part of the object. You end up re-executing the code to get at the data, change the data, and re-encapsulate the data with the code. You think that isn't overhead? In a non O O language, you read the data raw, in your own data layout (or existing file layout you reuse), mod the data, and write. An order of magnitude less machine use. (But you have to know your data! (i.e. you have to know what you are doing.))

You have to know what you're doing in an OO language, too.

And the object orientation is a feature. Data in an object is operated on by methods. Objects are accessed through an API. The whole idea is that you can change the methods (code) or the data the methods operate on, without any other part of the application knowing or caring. The object is a black box. The calling process issues calls in a particular format and expects to get results back in a particular format. How the object does it is the object's business. As long as call and response is unchanged, the calling process neither knows nor cares.

The big question is not how you change an object, but why you do so. And part of the skill required is understanding the problem you are trying to solve and designing objects that you are less likely to need to change down the road.

Java - an Interpreted language. Any interpretive language is an order of magnitude slower that a compiled language. (Sometimes more that one order of magnitude). A compiler writer is a special breed (I know just one.), who makes the compiled code efficient. Very efficient.

Not quite. Java compiles to bytecode. Bytecode is object code targeted at the Virtual Machine Java implements.

Other current languages like Python and Ruby use a similar design. The concept goes back to the UCSD P-system, which was used by the IBM Displaywriter and showed up in other places like the Apple II. You wrote in Pascal, but it compiled to P-code interpreted by the underlying P-system.

Garbage collection...Jeez, I thought that went out with BASIC in the early 1980's! IBM solved that problem back in 1975!!! (Had to! How else could you run 5,000 interactive terminals simultaneously on a machine the power of a Intel 486. One machine, 'cause it cost over a million dollars then!)

IBM didn't solve the problem for anything save a particular limited set of cases.

One of the biggest problems in programming has been the need to allocate memory used by your application, and free it again once done. In C, for example, you use malloc to do it, and it's on you to handle the case where there isn't enough memory available and fail gracefully. It's also on you to make sure the memory you allocated is properly freed when you're done with it, and "memory leaks" because this didn't happen are probably the single biggest headache for programmers. You did put in a "free" call to free the memory, but something happened and it never got executed....

In a language like Java, those details are handled for you. You can concentrate on the logic of the application, not the housekeeping required to make it work.

If you want to get rich as a programmer, figure out how to write a Java compiler that writes out non O O efficient code. Sell to the cloud world, where efficiency save the company money, unlike today's paradigm, where efficiency means nothing, because somebody else is paying for the hardware...

There have been tremendous advances in compiler technology since Java was first developed, and documented cases where a Java application runs faster than the same thing done in C.

Different languages evolved to solve different problems, and languages designed for a particular kind of problem get the nod as the best fit for the job. It's why we got things like Fortran, COBOL, BASIC, and C and very few people code in assembler any more.

Forget OO for a moment. Java has several compelling advantages.

1) It's cross platform. Java Virtual Machines exist for just about everything, including smartphones.

2) It's "write once, run anywhere." Java compiles to tokenized binary form called bytecode, targeted at the underlying virtual machine implemented by the JVM. The bytecode is the same, regardless of what you wrote it on. So you can take a well written Java program created on a Windows PC, compile to bytecode, and run that bytecode on a Mac, expecting it to look and behave the same.

3) Java runs by design in a sandbox, with a feature being that it can't easily affect the underlying system. You don't worry about rogue code running amok and doing Bad Things to the machine it's run on.

One of the things I'm playing with here is Eclipse, an IBM written programmer's IDE that they made open source and has contributions from developers around the world. Eclipse has a syntax highlighting editor, class browser, hooks to compilers, debuggers, libraries...everything a programmer needs to develop code. And it's extensible to handle new languages and tools.

It's written in Java. The same bytecode that implements it on my Windows PC is also running on my Linux based notebook. Just unpack the archive to the desired location and create a shortcut to run it.
______
Dennis

[DMcCunney]

Originally Posted by Worldwalker

TECO ... Text Editor & COrrector ... the editor I loved to hate, back in the day. What can I say, I liked EDT. Still do. My fingers could probably find the EDT hotkeys on a VT-100 even today, and it's been decades.

I logged time on EDT back when, and agreed, it was a potent editor. I have a couple of different EDT emulators here, and there's a commercial version I used to support under Unix at a previous employer.

This one is pretty decent open source version, written by a former DEC programmer, and available for a variety of platforms:
http://texteditors.org/cgi-bin/wiki.pl?SEDT

I used TECO and hated every minute of it.

Awww. Just because TECO code looked like line noise?

Though the fact that there was an Eliza game written in it was pretty awesome.

And the Eliza game wound up in Gnu Emacs, too...
______
Dennis

[mike_bike_kite]

Originally Posted by Worldwalker

Though the fact that there was an Eliza game written in it was pretty awesome.

I once wrote an Eliza program (in PHP this time) and called it the A.I. Wife - if you miss Eliza then feel free to have a chat with my version.

[nomesque]

*gibber*

Help! I'm remembering my university days! That's not supposed to happen, is it?

ARGH - flashbacks... computer lab... unix boxes... VI!!! Telnet!!! Linx! Eiffel!!!

*weeps and stuffs memories back where they came from*