MLX (software) - Commodore Versions of MLX

Commodore Versions of MLX

MLX was initially written for the Commodore 8-bit series of computers. It was first introduced in December 1983 for the Commodore 64. The following month, a VIC-20 version ("Tiny MLX") was introduced. Generally, MLX listings were reserved for relatively long machine language programs such as SpeedScript, while shorter programs were often printed in the form of BASIC loaders that used DATA statements to POKE the code into memory.

The initial Commodore versions of MLX accepted decimal data. Each line of data contained six data bytes, plus a seventh byte which served as a checksum. Although each line began with the memory address, and bytes were separated with commas, these did not have to be typed; the program auto-incremented the address and automatically printed the comma delimiters every three characters. Invalid keystrokes were filtered out. When a line was completely typed, it was automatically verified against the checksum value. If it matched, the user would be prompted for the next line. If not, a warning buzzer would sound and the invalid input would be discarded so that the user could retype the line. Several keyboard keys were redefined to create a makeshift numeric keypad.

Although the original MLX was a more efficient entry method than BASIC loaders with DATA statements, it had a number of issues. Foremost among them was that it POKEd the data entered directly into the affected area of memory. Since many machine language programs were located partially or entirely in the BASIC RAM area, this meant that the user had to manually adjust the pointers to the top and bottom of BASIC RAM (using POKE statements provided in the article) before running MLX to keep the entry program from encroaching on the code and vice versa. If a program was entered in multiple sittings, this had to be done each time. Also, decimal storage required seven keystrokes for each two bytes entered (taking into account the checksums). Furthermore, many transposition errors were overlooked by the original MLX's unsophisticated checksum algorithm.

Because of these issues, a completely new version of MLX (written by Technical Editor Ottis R. Cowper) debuted in the December 1985 issue of COMPUTE!. It appeared in COMPUTE!'s Gazette the following month. The new version stored typed data in a buffer, thus eliminating the need to reconfigure memory. The data format changed from decimal to hexadecimal, thus drastically reducing the number of keystrokes needed. Line length was increased from six bytes to eight, thereby reducing the checksum overhead. A more sophisticated hash algorithm was implemented to catch errors overlooked by the original, only failing to notice a substitution of 0xFF and 0x00 . This replacement version of MLX was very successful, and continued to be used until COMPUTE!'s Gazette switched over to a disk-only format in December 1993.