Fifth - virtual machine, operating system, programming language

1 !Project deprecated!

Current implementation does not support object oriented programming. While working on Fifth I got lots of cool new ideas that require reimplementation of everything.

Currently I try to implement those new ideas in the project called Sixth

System is built many years ago when I was still using DOS as a primary operating system.

2 Introduction

Fifth is programming lanquage & operating system, running on virtual CPU, with custom instruction set. It is much like Charles Chunk Moore's Forth, it also uses stack architecture, and many commands are similar. Basically I got familiar with concepts of Forth, and being inspired created my own system.


Read more about:

3 Installation

Just unpack all files, witout altering original directory structure, somewhere in your hard disk. For example: C:\MISC\FIFTH\…. To run fifth you need minimally just 2 files. emulator itself ( EMULATOR.EXE or EMULATOR.COM ), and virtual disk file ( DISK.RAW ).

Read more about distribution directory layout

4 Software/Hardware/Human requirements

4.1 Software:

  • MS-DOS 6.22, with HIMEM.SYS loaded.
  • Mouse driver if you have a mouse.
  • Does work only when CPU is in real mode.
  • To recompile ASM sources I used FASM (Flat Assembler).
  • I ran QBasic utilities on QB 4.5 .
  • VESA support by BIOS, or external driver (UNIVBE).

4.2 Hardware:

  • Minimum CPU 386.
  • 64 KB free RAM below 640KB,
  • 2 MB of free XMS.
  • VESA compatible video card.

4.3 Human:

  • Beginner level Forth knowledge is recommended.
  • Lots of enthusiasm.

5 Numbers representation


Because I can define everything, starting from CPU, why not try also alternative and unique number representation ?

Fifth uses its hexdecimal number representation as primary. Numbers shape is formed by dividing a square into four parts. And manipulating their color (black or white).

6 Disk file map, and it's data structures

Core and high-level boot code is stored outside of the filesystem to allow easy access to it, at early booting time, when filesystem is not yet initialized.

6.1 disk allocation

offset length description
0 ~4 Kb Fifth core
4 Kb ~32Kb high-level boot code
37 Kb ~65Kb FAT
101Kb ~16MB filesystem data area

6.2 FAT entry format:

code meaning
-2 last sector
-1 empty sector
0 -- .. pointer to next block

6.3 file entry format

offset length description
0 4 extension
4 16 name
20 4 entry point
24 4 size
28 4 last modification time

7 Core architecture

Fifth core is simply some amount of already compiled into machine code and linked together modules (entries in other words). In compilation process modules is compiled one by one and simply stored on top of already existing and growing core. Separately from core is kept dictionary, this is special list that contain names of compiled modules, variables etc. and they locations in core. Constants use dictionary space only. Random word can be removed from dictionary at any time. Currently dictionary can contain at most 1000 entries.

7.1 dictionary entry format

offset length description
0 4 0 < previous entry
    0 = last
    -1 = empty
4 15 module name string
19 1 entry type
20 4 entry data

Core headers as linked list of module names make up something like dictionary. When some entry address is needed compiler can quickly run through headers backwards and find needed entry.

7.2 Possible module types

type description "execute" action
0 data compile "num" instruction
    with address to module
1 submodule compile "call" instruction
    with address to module
2 imm. submodule immediately call to module

7.3 Memory map: (average)

<loc> <size> <desc>
0 ~4096 core
1500000 ~32000 highlevel Fifth boot code
200000h   core startup messages area
5200000   end of dynamic memory space

8 Fifth source format

Fifth uses a different character table and codes than ASCII (still almost similar). I call it FSCII (Fifth Standard Code for Information Interchange) for example space character is not 32 but 255 instead. I plan to use mainly HEX numbers, and create new characters to represent numeric values. So typical nemric characters "0123…" is treated like ordinary letters.

8.1 FSCII:

DEC HEX function
0 - 15 0 - F HEX numbers
252 FC backspace
253 FD tabulator (TAB)
254 FE carriage return (CR)
255 FF space
else   ordinary characters, same as in ASCII.