Just wanted to chime in with some historical perspective (no doubt with a few gaping holes, so feel free to correct me). There's a brief summary on the wiki of the prior discussions: http://www.haskell.org/haskellwiki/List_function_suggestions These are the two previous threads: http://www.haskell.org/pipermail/haskell-cafe/2006-July/thread.html#16559 http://www.haskell.org/pipermail/libraries/2004-July/thread.html#2342 Some interesting points: - the lines/words analogy to split/split': http://www.haskell.org/pipermail/libraries/2004-July/002352.html Which reinforces Gwern's point that the library should have both. - Perl, Python, and Ruby do not agree on split semantics: http://www.haskell.org/pipermail/libraries/2004-July/002351.html - there seems to be some desire for functions splitBy and tokens, which haven't been mentioned (yet) in this thread, I think. And finally my opinion: ByteString split:
split 'a' "aXaXaXa" == ["","X","X","X",""]
Neil's:
split' 'a' "aXaXaXa" == ["","X","X","X"]
I prefer the ByteString split (i.e. I prefer separator semantics over terminator), also because it's invertible. Some more support: http://www.haskell.org/pipermail/haskell-cafe/2006-July/016649.html Alistair