[Political Junkie Too]

In 1981, one of my first COBOL assignments was to rewrite a program to use VSAM files instead of ISAM files.

-PJ

[Myrddin]

One of my early challenges came in the form of data files with variable length records. It predated XML. I was given a 30 kilobyte example as a reference model to build a parser. Given a key, find the record and use the data. It tested fine. On a Saturday morning I received a call. The code was processing 9 records per hour. Humans can do far better. Ouch. The actual data was 10 files of 30 megabytes each. The sequential search was glacial. I added code to build an index using key:filename:offset:length and stored it in hashed access file (a UNIX library utility). Going forward, a key was presented, the index was fetched by key and the data fetched using the index record. Processing speed was increased from 9 records per hour to 112 records per minute. I didn't have language support e.g. ISAM vs VSAM style. It was just standard C on UNIX.