Eric Seidel
Eric Seidel
It's worth noting that `-fno-cse` doesn't help if the "config" type has strict fields, in which case the annotations are ignored again. For example, using the README's first example ```...
That does seem quite extreme, as you'd have to infer that foo f == g before adding the decreasing arguments constraint. On Wednesday, December 4, 2013, Niki Vazou wrote: >...
I would actually suggest splitting the `IO` example into a separate issue, I suspect it will be much more difficult to resolve than the negative-Mu business. I agree though that...
It might be nice to expose the date/time encoders as a separate package like you did with the parsers. Perhaps `iso8601-builder`?
Yes, I think that would work. But @ranjitjhala's suggestion yesterday was to do away with the pending substitutions altogether, which sounds much cleaner to me.
Yes, that sounds about right.
That seems to be the issue. We already have a nice error message for when we **can't find** z3, but we don't check whether z3 is working properly.
Something along the lines of: able to start-up and answer queries without crashing. As I understand it, in the referenced issue, `z3` was on the PATH and executable, but somehow...
Hrm, something's fishy here, upon closer inspection ``` λ> let e = (EApp "Set_mem" [EVar "x", EApp "keys" [EVar "y"]]) e :: Expr λ> showpp e "(Set_mem x (keys y))"...
@spinda thanks for digging around! I'll look into a fix next week.