Automated Code Transformations: Dealing with the Aftermath

Dealing with legacy systems has been a challenge for the industry for decades. The pressure to efficiently modernise legacy assets to meet new business needs and minimise associated risks is increasing. Automated code transformation, which is associated with serious (long-known) risks, is a high priority in the industrial environment due to the cost structure, the effort required and the supposed time savings. However, little has been published about the long-term effects of successful migrations. This paper looks at three different cases of automated code transformation at different stages of their lifecycle, highlights the lessons learned and derives a number of recommendations that should be useful for planning and executing future transformations.

I. Introduction

Van Deursen et al. observed in 1999 that we, as an industry, are sitting on a “software volcano” [1]. This was the time when the change of the century (Y2K problems) as well as the Euro conversion in Europe forced us to take a detailed look at our existing code bases. In 2000 Terekhov et al. [2] observed that managers sitting on a large pile of legacy code are susceptible to “silver bullet” [3] solutions recommending the easy and quick automated translation of legacy code, however thin the proof for actual existence and sustainability [4] of such a solution might be. He called this “name magic” and concludes with “Easy conversion is an oxymoron”, seeing “much work must be done before language converters give satisfactory results on real-world code.”