A Brief History of the Pick Environment in Australia - History of Computing: Learning from the Past - page 149

needed, each one equivalent to a table. The database was then all the files that were
related. Typically one would create an account called “Student Records” and all the
related files would be stored there.
Pick differs from other database models because it allows fields to have repeating
values and for one field to be a “controlling” field with others defined as “dependent”
fields. This allows synchronization of repeating fields. It is also possible for any of
these individual repeating fields to themselves store repeating “sub-fields”.
Whereas people using the relational model must normalize their structures, the
Pick model allows avoiding first normal form and the consequential join tables. Look
at the example of a student database that stores student data as well as multiple sub-
jects studied and as well as results for multiple assignments for each subject.
The repeating fields at 12, 13, 14 and 15 break Codd's 1NF law but allow this re-
peating data to be stored inside the original record instead of creating an additional
table that will need to be 'joined back' later when accessed. Attribute 15 holds repeat-
ing sub-values aligned to the controlling attribute 12. Here we are able to model
assignment results for a given subject where there might be a varying number of as-
signments for each subject and even varied according to the semester taken.
Fig. 2. Example of the Actual Data Record on Disk that is Read and Written
Thus one of the key differentiators of the Pick data model is that each “master re-
cord” can also contain all the detail elements associated with transactions on that re-
cord. In the example shown, the one structure has replaced several relational tables, or
indeed several network-based DBMS detail files. This is quite a saving in not only
potential disk space but more importantly in the disk I/O necessary to perform queries
and updates.
2.2 The Operating System Model
There was a stranglehold by the major vendors on their proprietary hardware and at-
tached host operating systems. IBM, HP and DEC were but a few of the major play-
ers. It is now of historical interest the difficulties that IBM found themselves in with
OS360 described by Brooks [5]. The Pick operating system was multi-user and time-
sharing with the ability to run dozens of serial users on an Intel 486 computer with