[Isaac, you need to join the TH list so you can post to it without me having to intervene.]
Done
Your original makeSum looked better.
OK, thats what I was hoping.
Solution: use the "smart constructors" from the paper. More like this:
genSummers n = fun ("makeSum" ++ (show i)) [clause [] (normal (makeSum i)) [] ]
Thanks, for some reason I was blind to the combination of using the "smart constructors" in together with sequenceQ. I thought I had tried every combination to make the types work out. Here's my solution below. peace, isaac \begin{code} -- |Given a size, n and n expressions, sum them up, so: -- > $(makeSum 4) 1 2 3 10 -- => 16 -- > $(makeSum 4) 1.0 2.2 3.4 10.1 -- => 16.7 makeSum :: Int -> ExpQ makeSum n = lam aPats (sumExps aVars) where -- create the symbols we'll need: aPats ::[Pat] aVars ::[ExpQ] (aPats,aVars) = genPE "num" n sumExps :: [ExpQ] -> ExpQ sumExps (h:t) = [| $h + $(sumExps t) |] sumExps [] = [| 0 |] -- |Bring functions named makeSumX into scope where X is in [1..n] genSummers :: Int -> Q [Dec] genSummers n = sequenceQ [fun ("makeSum" ++ (show i)) [clause [] (normal (makeSum i)) [] ] | i <- [1..n] ] -- Then in a different file, I can say -- $(genSummers 10) -- to bring makeSum1, makeSum2, etc into scope \end{code}