Haskell High Performance Programming

Handling sequential data

The standard list,[], is the most used data structure for sequential data. It has reasonable performance, but when processing multiple small values, say Chars, the overhead of a linked list might be too much. Often, the convenient nature of [] is convincing enough.

The wide range of list functions in Data.List are hand-optimized and many are subject to fusion. List fusion, as it is currently implemented using the foldr/build fusion transformation, is subtly different from stream fusion employed in ByteString and Text (concatMap is a bit problematic with traditional stream fusion). Still, the end result is pretty much the same; in a long pipeline of list functions, intermediate lists will usually not be constructed.

Say we ...