RSEConUK 2019

How do we define legacy software?
Code that falls into this category can be typically classified by a lack of tests, or use of outdated practices, including unsupported dependencies, or having a code base written in an obsolete programming language. Whilst this list is not exhaustive, these are usually the signs that software is going to cause you issues when it comes to attempting to extend or scale the code.

Characterisation of code then falls into two simple categories, legacy code, and not legacy code. With no clear definition on legacy software, the boundary between the two is far from clear, and is constantly moving. However, one thing that remains clear and consistent is that every piece of software will encounter the issue of migrating legacy systems in its life cycle.

Generally, there are three possible solutions when dealing with legacy software: refactor, rewrite, or retire. All three options should always be considered when reaching that point in the software development life cycle, and this talk examines the costs and benefits for each. A strong focus on testing is leveraged in this talk when considering either rewriting and refactoring and establishing a good set of tests is paramount when using these approaches.