Jonathan -
Lots of good points. Let me try and address some of them.
Why should the user have to explicitly specify a list of contexts in a
separate file -- why couldn't contexts just be inferred from those the
user is already using, like in FMP? They could of course be cached in
a .context_cache file for faster access, but it seems to me that
having an explicit, separate configuration file is entirely
unnecessary.
The other complexity I don't yet fully grok is why there is this
notion of task-type that's orthogonal to both context and project.
>From my own experience, at least, this sort of added layer of
complexity over and above basic GTD (traditionally you just have
@email, @work and @home, not tasks vs. email vs. ? | @home vs. @work
leading to email@home vs. task@work) always winds up getting trimmed
away in the end, and, where the distinction is genuinely useful, it's
usually because you actually have a new context distinction
(@work-email and @home-email), not because these sorts of things need
to be fundamentally orthogonal throughout, complexifying the system
even in cases where the location is irrelevant.
Anyway, just my 2c after watching from a distance for a while. I'm
quite interested to see where this goes.