As difficult to maintain legacy code bases proliferate within an organisation, tackling them head on often requires a substantial engineering investment without any guarantee of success.
We examine such a code base from within our own organisation - a large, stateful, object orientated .NET API for distributed computation with little type-safety.
It became apparent that idiomatic use of this library encouraged bad coding practices amongst its users, making it harder to make meaningful improvements to the API, and cementing its position as a legacy component.
When faced with this problem, we eschewed a big rewrite or major refactor in favour of abstracting over the problematic design with a declarative, type-safe facade written in F#.
This talk explores the significant advantages this approach offered to both the library’s end-users and its development team - immediately preventing further accumulation of problematic user code, and establishing a better conceptual model for the library’s functionality while laying a pathway for the ultimate replacement of the legacy implementation.