Club Amiga de Montreal - CAM/CAM_CD_1.iso/files/032.lha/doc/root directory/WhyDraco

Text File | 1987-05-17 | 8.1 KB | 191 lines

    What is Draco, Why Did I Write It and Why Is It Like It Is?

    I usually describe Draco (pronounced Dray-ko) as a "systems programming
language". That means that it is a language which is suitable for what I
think of as systems programming - writing operating systems, compilers,
editors, databases, etc. This doesn't mean that it isn't suitable for other
applications such as writing games, graphics programs, numerical programs,
etc. It does mean that the language has all of the facilities needed for
the former type of programming, such as bit operators, pointer
manipulation, support for complex data structures, etc.

    What is different about Draco? I won't try to compare it with every
other programming language in the world; instead I'll stick to two of the
most popular ones for micros nowadays - C and Pascal. Draco has all of the
facilities of C, except for bitfields and the macro preprocessor. Unlike C,
and like Pascal, it is a strongly typed language. This means that it won't
let you assign an integer to a pointer (unless you really insist). It is
also not an expression language like C, thus it makes a quite strict
distinction between statements like "a := 27" and expressions like "a +
27". Pascal is strongly typed, but lacks many of C's facilities - pointer
manipulation, bit manipulation, standard separate compilation, conditional
compilation, etc. I like to think that Draco combines the best features of
both languages.

    Visually, Draco doesn't really resemble either language closely, but is
a little closer to Pascal than C. It uses ':=' for assignment and '=' for
comparison, like Pascal and unlike C. It's structure and union declarations
are like those of C, however. As a simple comparison, here follows the same
program, written in Pascal, C, and Draco:

Pascal:

    PROGRAM test(INPUT, OUTPUT);
    VAR
    i, j : INTEGER;

    BEGIN
    FOR i := 0 TO 10 DO BEGIN
        FOR j := 0 TO i DO
        WRITE(j : 2, ' ');
        WRITELN
    END;
    WRITELN("All done.")
    END.

C:

    #include <stdio.h>

    main(argc, argv)
    int argc;
    char *argv[];
    {
    int i, j;

    for (i = 0; i <= 10; ++i) {
        for (j = 0; j <= i; ++j)
        printf("%2d ", j);
        printf("\n");
    }
    printf("All done.\n");
    }

Draco:

    proc main()void:
    int i, j;

    for i from 0 upto 10 do
        for j from 0 upto i do
        write(j : 2, ' ');
        od;
        writeln();
    od;
    writeln("All done.");
    corp;

    First, it's clear that the C program has lots of brackets, while the
Pascal and Draco programs have lots of keywords. A significant difference,
not very clear in this small example, is that Draco uses different keywords
for each job, rather than relying on a single construct (the BEGIN - END or
'{' - '}' block). I greatly prefer keywords, finding them easier on the
eye. I also prefer languages in which the use of case (UPPER v.s. lower) is
available for my own purposes, rather than having them equivalent as in
most Pascals. Also, note that the Draco program uses 'upto' in the 'for'
loops - this tells the compiler that the loop will be counting upwards;
'downto' is used for downward counting loops. C doesn't really have a
semantically different 'for' loop - it's just a kind of shorthand for a
'while' loop.

    Some of the inadequacies of Pascal from my point of view are as
follows:

    - no standard separate compilation
    - no conditional compilation
    - no general string mechanism
    - no pointer manipulation
    - no bit manipulation
    - I HATE BEGIN and END!
    - no signed/unsigned types
    - limitations on function and argument types
    - procedure calls don't use '()' - they look like variables
    - no typed, named, constants
    - no available, decent implementations (fast compilation, good
        code, nice libraries, good error reporting)
    - I/O semantics that are poor for interactive programs
    - no file inclusion or module specification facility

    Some of the inadequacies of C from my point of view:

    - too many bloody brackets!
    - horrible declaration syntax (just what is "char *(*p[])()"?)
    - error prone conventions (how many times have YOU written '='
        when you meant '=='?)
    - non-portable I/O (if you don't believe this, take a look at the
        open calls on CP/M or MS-DOS versions of C, where you get to
        tell it what it's supposed to do with '\n')
    - potential for extremely unreadable code (misuse of macros, etc.)
    - slow compilers (as I've heard it, the reason that the original
        UNIX C compiler for the PDP-11 generated assembler source was
        so that the compiler writers didn't have to worry about long/
        short branch optimization, since that was done by the
        assembler. Producing assembler source is just plain slow. Those
        who argue that they want to hand edit it to improve it are
        crazy!)
    - lack of much type checking (I prefer compilers that tell me about
        my dumb mistakes. This is a lot better in the ANSI draft
        version.)
    - inefficient standard setup - passing everything as 16 bits on an
        8 bit CPU isn't so hot (or 32 bits on a 16 bit one)
    - stupid linkers - why add all that code I'm never calling?
    - no built-in I/O - this makes even the simplest programs large
    - no typed, named constants

    All of these issues have been addressed in the Draco language and
tools. Just as important to me is the quality of the tools (compiler,
linker, etc.) The Draco compiler goes from source code to relocatable,
optimized machine code at a rate of about 2000 lines per minute on a 5 MHz
8080. Working from one 8" floppy disk, the entire compiler (about 10,000
lines) can be rebuilt in under 10 minutes. No other compiler I've heard of
for CP/M can do this (at least not and produce good code). The linker will
link small programs in one quick pass, and won't load any code that isn't
referenced by the program. A simple "hello there world" program is under
1000 bytes. The Amiga version of the compiler operates at about the same
speed, and the linker I'm using (BLink) isn't bad, but doesn't selectively
load like mine does.

    Another reason that these programs exist is that I LIKE writing
compilers and stuff. I'm up to about seven compilers now, the latest of
which is a C compiler that should meet the ANSI draft standard (it's a huge
monster written in C, but at least I was paid to write it!)

    So I've written my very own personal compiler, that does things just
the way I want; why should anyone else want to use it? Put simply, the
Draco package (which includes the various libraries I've built up) is
possibly the most effective way to produce compact, efficient code for CP/M
systems. In the past couple of years, asside from fine tuning the compiler,
I've written somewhere around 20,000 lines of Draco code, including the
screen editor I'm typing this into, a complex graphic role-playing game,
several CRT-oriented games for CP/M, ranging from the trivial to the quite
complex, a database package, a text processor (with a friend), a modem
program, etc. If you want to program a CP/M system, whether for fun, profit
or whatever, and are willing to learn another language, then I feel that
Draco is a valid choice. The Amiga version hasn't been used as much yet,
but appears to be reasonably solid. I plan on bringing it up to the
standard I set with the CP/M-80 version, but this will take time and a lot
of work.

    To be fair, I will end this intro with a list of things that I find
are lacking in this version of Draco:

    - essentially non-existant floating point support
    - no proper modules (although Draco goes about half way to
        providing a usable kind of module)
    - no bit oriented type (I haven't yet fully convinced myself that
        this is needed)
    - error handling is considerably better than most C compilers I've
        heard of, but it could still use some improvement
    - object code can ALWAYS use improvement, but the improvements
        that are left would either be difficult or of little actual
        benefit and would probably make the compiler too big to fit on
        standard CP/M systems. This isn't as much of a problem on the
        Amiga, but I do plan on keeping the compiler small enough to
        run well on a 512K machine.

and, of course

    - Draco is supported only by me, and available only on the systems
        that I choose to put it on (currently CP/M-80 and Commodore
        Amiga)