Converting existing code into SpaDES

The previous parts of this book taught you how to build SpaDES workflows – modules, projects, experiments – using the conventions that make them composable, reproducible, and re-runnable. But most real projects don’t start clean: there is already a working analysis sitting in an ordinary R script. The question is how to bring that existing code into the SpaDES ecosystem without rewriting it from scratch. Once it is in, it inherits all the properties the framework provides for free: caching, scheduling, swappable parameters, metadata- driven composition with other modules.

This final part walks through that conversion twice, on the same example – a caribou RSF forecast script. The first chapter does it by hand, step by step, so the mechanics are clear: scaffold the module with SpaDES.core::newModule(), fill in the metadata, then translate the linear setup section into an init event and the annual loop into recurring events. The second chapter does the same conversion as a short conversation with an AI assistant – the metadata block, the event functions, and the debugging cycle are exactly the kind of boilerplate-heavy, convention-bound work an LLM handles well. Both paths land at the same RSFpredict.R; the AI version is just faster and is how new modules are likely to be authored from here on.