Software

We've now outlined the concepts and processes of questionnaire management, and shown that their interpretation within the current system does could be improved in some key areas:

• the models of selections, parameters, structure, constraints, and layout can all be generalised for greater coverage of multi-tenant requirements.

  selections cannot be defined over multiple datasets, their structure has a fixed depth, it must be homogeneously composed and cannot be divorced from layout; parameters are fixed; constraints are restricted in a number of ways; the language of layout is limited, both in terms components and in terms of their configurability.

• processes could be at once more explicit and flexible.

  Campaigns are not directly represented and there are no facilities for bulk generation, publication, and ingestion; specifications effectively remain internals of the system, they are not exposed to administrative users and there's no dedicated tooling to manage them; the lifecycle of specifications and questionnaires is largely implicit, hence opaque to monitoring and assessment activities; publication is exclusively offline.

We discuss now how the system will evolve to address its current shortcomings. We look first at the architecture of the solution, then to the key points of its design.