Tom Smeding

Another example of similar behaviour, except now there is no dead code, but there is a more complicated scalar expression to reason about. ```haskell let a = use (fromList (Z...

Two more programs that produce (somewhat) unexpected results: ```haskell zipWith (+) (zipWith (\x y -> x * 2 * y) a a) (map (\x -> x * x) a) ```...

```haskell map (\a -> let b = a * a ; c = b * b in T2 c (T3 a b c)) (use (fromList Z [1 :: Float])) ```...

**EDIT: please see my next message for a smaller repro** This one I haven't been able to reproduce with fusion-processed programs, only on the so-called "internal AST". Therefore, a ghci...

Smaller reproducing example for the above missing-parentheses issue: ```haskell T2 (T2 (lift ()) (use (fromList Z [1]))) (use (fromList Z [2])) :: Acc (((), Scalar Float), Scalar Float) ``` `show`s...

Can I petition for a mode in the pretty-printer that attempts to output valid Haskell input code? Having to manually translate a zillion occurrences of Tn to In is... time-consuming.

Another case where putting multiple things on a single line produces ambiguity: (EDIT: actually, more likely to be a precedence issue somewhere) ```haskell let a = use (fromList (Z :....

Pretty-printing a single-element tuple is inconsistent between left-hand sides and right-hand sides. This is due to the special cases in [prettyAtuple](https://github.com/AccelerateHS/accelerate/blob/af73d11ede9d947029b0257398d2420b05901364/src/Data/Array/Accelerate/Pretty/Print.hs#L275) and [prettyTuple](https://github.com/AccelerateHS/accelerate/blob/af73d11ede9d947029b0257398d2420b05901364/src/Data/Array/Accelerate/Pretty/Print.hs#L508) which are not treated specially in [prettyLhs](https://github.com/AccelerateHS/accelerate/blob/af73d11ede9d947029b0257398d2420b05901364/src/Data/Array/Accelerate/Pretty/Print.hs#L296)....

@dschrempf Firstly make sure that what you want is actually possible to do using Accelerate. Accelerate has its own notion of arrays; there is [accelerate-io-vector](https://hackage.haskell.org/package/accelerate-io-vector) which @tmcdonell may be able...

In the force-push: - Repeated `strcmp` on `"--perms"` and `"--size"` factored out to `is_modifier_option()` - Somewhat more self-contained commits Parsing of tmpfs sizes is not yet done in bwrap, see...