Program understanding using program slivers-an experience report
Aarti Gupta · 1997
A large amount of legacy software is written in languages such as FORTRAN and COBOL. Hard-copy and inline documentation for this software is often inadequate to perform maintenance or re-implementation. These tasks are further hampered by legacy structures and programming style such as the use of “goto”. This paper describes our experience using simple pattern matching tools to obtain “slivers” from legacy process information systems. These slivers provided information that assisted in recovering program design. These slivers also helped in identifying segments which could cause problems with two-digit year fields at the turn of the century