Sarah Jelinek wrote: > Hi Jean, > > Thank you for the review. My comments inline.. > > >> Section 2.1: >> >> "The engine is capable of beginning execution at the next unreached >> checkpoint in the sequence" >> >> I find that confusing. If you haven't reached that checkpoint, how >> can you resume execution at it? > > we aren't really resuming. we are starting that checkpoint at its > beginning. > >> (An aside, probably should change begin to resume) > > I think begin is better than resume. The idea is that the checkpoints > can be run independently(within sequence of course), and that the > engine can be told which checkpoint to begin execution at. > > >> Under checkpoint methods: >> You might consider methods such as resume, pause, rollback. Better >> names could be selected but I suspect you get the idea. > > I am not sure we need those. I get the idea, but I am not clear why > you think we need them. A checkpoint must be run in its entirety, and > the engine is responsible for managing what has been run and where it > needs to start next. So, the checkpoint itself doesn't manage that. > Can you clarify why you think these are required? Could be we haven't > been clear enough in our description. I think I'm just coming in with to many preconceived ideas of what checkpoints are from the DC work. If no one else is bugged by any of this I'll be fine.
Jean
