Towards a Safe Method for
Computing Dependencies in Database-Intensive Systems
Csaba Nagy, János Pántos, Tamás Gergely and
Determining dependencies between different components of an
application is useful in lots of applications (e.g., architecture
reconstruction, reverse engineering, regression test case selection,
change impact analysis). However, implementing automated methods to
recover dependencies has many challenges, particularly in systems using
databases, where dependencies may arise via database access.
Furthermore, it is especially hard to find safe techniques (which do
not omit any important dependency) that are applicable to large and
complex systems at the same time. We propose two techniques that can
cope with these problems in most situations. These methods compute
dependencies between procedures or database tables, and they are based
on the simultaneous static analysis of the source code, the database
schema and the SQL instructions. In this paper, we quantitatively and
qualitatively evaluate the methods on real-life data, and also evaluate
them on some of their potential applications.
dependencies, program analysis, databases, SEA relations, CRUD matrix.