I recently purchased an assembler for my NEC PC-8201A which included
a program to determine RAM file sizes. This program was an example
of why I purchased an assembler in the first place. I soon discovered
a small quirk in the program's logic with the BASIC files. Using the
companion interactive debugger, I discovered that Microsoft encoded
their BASIC line statements as follows:

        Next Line   Current Line
         Address       Value

         LSB  MSB    LSB  MSB     TEXT & TOKENS

The programs original theory was correct because Microsoft did encode
the end of each BASIC program with two hex 00 bytes, but ther are other
occurances of the double hex 00 bytes within a BASIC program.  If you
number your BASIC program beginning with line number zero, the example
program would return a length of 4 bytes.  There are also other ways
that this can occur because of the way constants are stored in the
tokenized line statement.

From the BASIC statement encoding above, I noticed that a program could
go from one statement to the next by repeatedly loading the next line
address and jumping to that location.  When this next line address was
zero the program had indeed found the actual end of the file.  By sub-
tracting the starting address from this ending address, a program could
calculate the actual length of the file and would work even if the first
statement's line number was zero and it's probably faster than counting
each character in the file.

I hope that you enjoy this enhancment.  You can reach me at the 
following if you would like to leave any comments:

     Computer Insights
     1910 Knob Creek Rd.
     P.O. Box 2466
     Johnson City, TN 37601

     SOURCE:     BCN041
     Delphi:     CLAWS
     MCI:        247-7899
     Compuserve: 74646,656

Sincerely,
Chuck Laws