A verification tool developeris vade mecum

The Edinburgh Concurrency Workbench has been the author's responsibility for the past four years, having been under development for eight years before that. Over its lifetime, we have learnt many lessons and encountered many questions about verification tool development, both from bitter experience and from discussion with other tool developers and users. In this paper, I raise some of these questions and give some personal comments on them. Some if these issues arise more generally, in other tools which are formally based or developed in academic environments or both.