reChunk is implemented like this:
reChunk :: [ByteString] -> [ByteString]
reChunk [] = []
reChunk (c:cs) = case B.length c `divMod` 2 of
(_, 0) -> c : reChunk cs
(n, _) -> case B.splitAt (n * 2) c of
~(m, q) -> m : cont_ q cs
where
cont_ q [] = [q]
cont_ q (a:as) = case B.splitAt 1 a of
~(x, y) -> let q' = B.append q x
in if B.length q' == 2
then
let as' = if B.null y then as else y:as
in q' : reChunk as'
else cont_ q' asIn the main body of the function, there is no check for empty chunks; those just pass through. In the cont_ worker, however, there's some effort (the B.null y check) to avoid producing an empty chunk. Why this apparent inconsistency? I would expect to either check for empty chunks both places or neither.
reChunkis implemented like this:In the main body of the function, there is no check for empty chunks; those just pass through. In the
cont_worker, however, there's some effort (theB.null ycheck) to avoid producing an empty chunk. Why this apparent inconsistency? I would expect to either check for empty chunks both places or neither.