A Context for Requirements Handling in an Agile Workflow
Abstract
When King’s Digital Lab (KDL), an RSE unit based in the Faculty of Arts and Humanities at King’s College London, began operations in 2016 we quickly adopted the industry-standard DSDM Agile methodology. Its emphasis on a fixed-time-and-cost basis with flexibility on features aligns well with the grant-funded projects work we commonly do. Our SDLC includes DSDM-recommended steps and documentation which we have found very helpful. However, because we have small solution development (SDT) teams each working on several projects simultaneously, workflows can be complex and it has been challenging to fully and consistently follow Agile precepts, especially with respect to handling project requirements. We like the theory of moving from initial high-level requirement descriptions to greater specificity via iterative development but in practice have found that while a typical Agile try/asses/adjust/try again fosters creativity and refinement in implementation it also brings a constant risk of confusion and inefficiency for the SDT. We have come to see that fast-paced iteration demands a fixed context: something consistent, intuitive, and convenient that allows both SDT and project partners to produce and keep track of evolving requirement descriptions and the constantly moving parts - discussions, decisions, task creation, etc. - that surround them. ‘Context’ here is not quite an approach (as, for instance, Specification by Example), or a formalized set of practices (like the Requirements Development & Management Practice Area of CMMI), or a purpose-built software tool (e,g. Altium 365) but might incorporate aspects from any or all those things. Drawing from our experience at KDL this talk will firstly discuss specific issues in requirements handling which can thwart Agility and frustrate all parties in a project. It will then set out aims, principles and practical considerations pertinent to a fixed context that facilitates rapid iteration in an environment of multiple complex workflows. Acknowledgements Colleagues at King's Digital Lab, King's College London, and in particular Senior Research Software Engineer Geoffroy Noël with whom I have had many helpful conversations on the subject of the talk.A recording of this session is available on YouTube: https://youtu.be/O27tomTwXD0