RPGDX > Forums > Design and programming > What follows Objects?: The Next Generation of Programming

tcaudilllg

I've produced this:

http://unitedinfinity.atspace.com/Projects/JS/Languages/StaticVariableJS.html

And I'm working to finish a simple programming language.

http://unitedinfinity.atspace.com/Projects/JS/Languages/DyneBasic.html

I aim first to replicate the modern icons of programming functionality in an advanced version of this "Dyne" language. These functions will include arrays, and objects. I will create special object structures, also, called "fields" that are timed to execute a self-contained list of instructions using information native to themselves at regular intervals, as like magnetic or gravitational field apply force to all objects within their respective ranges at a constant rate. I will include variables called "tensors" whose value may only be a solution of a predefined equation. With these I believe I can effectively model a virtual universe.

Is there something else beyond this? What else is there? Is the science of programming as far now as it can go, do you think? Or are there unheard of new ways of thinking lying just around the corner...?

LeoDraco · Posted: Fri Jan 21, 2005 2:00 am Post subject:

Galbalan, honey, I just love how you post links to things without any documentation towards how to utilize them. That, as is said, is teh sexy.

EDIT: Also, there is this thing called The World Wide Web Consortium, right? They have this validation service, which, one could intuit, validates SGML-based languages. A sample validation of your StaticVariableJS document should enlighten you that your utilization of HTML is over six years out of date.
_________________
"...LeoDraco is a pompus git..." -- Mandrake

biggerUniverse · Posted: Fri Jan 21, 2005 8:04 am Post subject:

And anyway, apparently the Next Big Thing(tm) after OOP is AOP (ugh).
_________________
We are on the outer reaches of someone else's universe.

tcaudilllg · Posted: Sat Jan 22, 2005 1:52 am Post subject:

http://www.livejournal.com/users/lordgalbalan/46896.html for the documentation.

I thought object oriented programming already did that (AOP).... Javascript does, at least. :P

tsb · I wanna be a ballerina! Joined: 09 Oct 2004 Posts: 23

Aspect-oriented programming is not the "next big thing", nor does JS have any direct support for it. JavaScript's reflection mechanisms are sufficiently powerful to effect aspects as a library, but I am unaware of any such. Java programmers can look at AspectJ for aspects. I believe there is a similar library for Python, but it escapes me.

Anyway, AOP is kind of cool, but it doesn't look to be very useful outside a very small, specific type of application domain.

I rather expect functional programming languages to start gaining momentum, as CPU speed is starting to cap out. Before long, the only way to improve performance will be to take advantage of multiple processors, and functional languages rock the crap out of C++ for that. :)
_________________
—andy
http://ika.sf.net

biggerUniverse · Posted: Sun Jan 23, 2005 9:51 am Post subject:

I agree that AOP is not general useful, but "The Next Big Thing" is always a hype anyway.

janus · Mage Joined: 29 Jun 2002 Posts: 464 Location: Issaquah, WA

biggerUniverse · Posted: Mon Jan 24, 2005 12:14 am Post subject:

tsb · I wanna be a ballerina! Joined: 09 Oct 2004 Posts: 23

Hajo · Posted: Mon Jan 24, 2005 10:20 am Post subject:

tsb · I wanna be a ballerina! Joined: 09 Oct 2004 Posts: 23

Hajo · Posted: Tue Jan 25, 2005 10:41 am Post subject:

Right. I think AOP is in the early phase of an emerging idea. We don't know what it will be good for, notr do we know the dangers or the benefits in detail.

It seems to address a gap in the current methodologies. If it will be an addon, or if a new methodololy will be created upon AOP is unknown.

I think it is a very intersting idea. It addresses some problems that disturb me since a long time. I'm an OO developer, but most of my objects are heavily interlinked and less ress reusaebal because each has to care about aspects that are not core functionaliyt fo the component, like error reporting or data serialization.

Maybe AOP is only part of the next big thing, the first glance that we see rising on the current knowledge horizon. But it seems to address important questions, and I hope the researches will porgress quickly and give us new ideas how to improve our code.

ATM my interest in AOP is only of theoretical nature. But since it adresses questions that I face rather often, I'm very interested in the solution it provides or will provide. Currently I think the solutions are not of production quality, but it looks very promising to me.

tcaudilllg · Posted: Thu Jan 27, 2005 1:52 am Post subject:

I believe the Dyne beta is functional.

You may want to experiement with it. It has several commands.

"vio" (virtual input/output) assigns a value to DHTML property or executes a DHTML method.

"if" does what it says. (ah, I forgot to add "else"....) Open with "{", terminate with "}".

"goto" goes to a different line. (syntax "goto [value]")

"line" changes a source line. (syntax "line [index] = [value]")

"name" changes the name of a variable. ("name [varID] = [value]")

Variable initialization is automatic, BASIC-style. Assign a value to a variable to create it.

For any case where a value can be used, a string variable can be substituted.

All tokens must be seperated by a space. Line terminators are semicolons.

!Updated:
Added support for user-defined types to Dyne. Use the create command ("create [value] : [varID]") where value is the template for the type.

Type templates are formed with the syntax "[type name] { [property1] [property2] [...] }". Type templates can be nested, so you can make a type that has a property has a property, like this:

"NewType { PropertyA { No1 No2 } }"

This creates four distinct variables: "NewType", "NewType.PropertyA", "NewType.PropertyA.No1", and "NewType.PropertyA.No2".

Using the create statement, we have:
[code]
Var = 0;
Type = "NewType { PropertyA { No1 No2 } }";
create Type : Var;
[/code]

This program produces a total of five variables: Var, Var.NewType, Var.NewType.PropertyA, Var.NewType.PropertyA.No1, and Var.NewType.PropertyA.No2.

Dyne's type support is still kinda shaky, because it seems to be having difficulty discerning between strings and variables, and it's making variables named after strings when it should be leaving them alone.... For best results, only use types with one property for now. Note that you can make a property with the create statement, and then add more properties onto it later. Dyne sees only variables, and what goes for one kind of variable goes for all of them.

[code]
TypeA = "PropertyLevel1";
create TypeA : Var;
TypeB = "PropertyLevel2";
create TypeB : Var.PropertyLevel1
[/code]

This creates Var, Var.PropertyLevel1, and Var.PropertyLevel1.PropertyLevel2, respectively.

...Yeah I fixed that object bug. Objects work peachily now. (or types or structs or whatever)

tcaudilllg · Posted: Fri Jan 28, 2005 9:25 pm Post subject:

I don't know much about functional programming, but it sounds really complex and hard to imagine....

Bjorn · Posted: Sat Jan 29, 2005 10:44 am Post subject:

Actually at my university, functional programming was the first kind of programming we were tought. It is very easy to understand and many problems, especially of recursive or mathematical nature, can be described in a short way that makes sense.

An example of a functional language, and the one I learned, is Haskell. In particular, take a look at the introduction to Haskell in which they compare for example a functional (Haskell) implementation of quicksort with an imperative (C) implementation, and compare the differences between functional and imperative programming in general.

There is a free interpreter for Haskell that's called Hugs.