Commodore 1541-II - Writing Relative File Data; Designing a Relative Record

WRITING RELATIVE FILE DATA
The commands used to read and write relative file data are the same Print#, Input#,
and Get# commands used in the preceding chapter on Sequential files. Each command is
used as described there. However, some aspects of relative file access do differ from
sequential file programming, and we will cover those differences here.
DESIGNING A RELATIVE RECORD
As stated earlier in this chapter, each relative record has a fixed length, including all
special characters. Within that fixed length, there are two popular ways to organize
various individual fields of information. One is free-format, with individual fields varying
in length from record to record, and each field separated from the next by a carriage return
character (each of which does take up 1 character space in the record). The other approach
is to use fixed-length fields, that may or may not be separated by carriage returns. If fixed
length fields are not all separated by carriage returns, you will either need to be sure a
carriage return is included within each 88 character portion of the record. If this is not
done, you will have to use the Get# command to read the record, at a significant cost in
speed.
Relative records of 88 or fewer characters, or final portions of records that are 88 or
fewer characters in length, need not end in a carriage return. The 1541 is smart enough to
recognize the end of a relative record even without a finahcarriage return. Though the
saving of a single character isn't much, when multiplied by the number of records on a
diskette, the savings could be significant.
Since each relative record must be written by a single Print# statement, the recom
mended approach is to build a copy of the current record in memory before writing it to
disk. It can be collected into a single string variable with the help of Basic's many string-
handling functions, and then all written out at once from that variable.
Here is an example. If we are writing a 4-line mail label, consisting of 4 fields named
"NAME", "STREET", "CITY & STATE", and "ZIP CODE", and have a total record
size of 87 characters, we can organize it in either of two ways:
WITH FIXED LENGTH FIELDS WITH VARIABLE LENGTH FIELDS
Field Name Length Field Name Length
NAME 27 characters NAME 31 characters
STREET 27 characters STREET 31 characters
CITY & STATE 23 characters CITY & STATE 26 characters
ZIP CODE 10 characters ZIP CODE 11 characters
Total length * 87 characters Potential length 99 characters
Edited length 87 characters
With fixed length records, the field lengths add up to exactly the record length. Since
the total length is just within the Input buffer size limitation, no carriage return characters
are needed. With variable length records, we can take advantage of the variability of
actual address lengths. While one name contains 27 letters, another may have only 15,
and the same variability exists in Street and City lengths. Although variable length records
lose 1 character per field for carriage returns, they can take advantage of the difference
61