List function suggestions

Let's fix this

We need these useful functions in Data.List; I'll call them 'split' (and variants) and 'replace'. These are easily implemented but everyone always reinvents them. The goal is clarity/uniformity (everyone uses them widely and recognizes them) and portability (I don't have to keep reimplementing these or copying that one file UsefulMissingFunctions.hs).

Use this page to record consensus as reached on the Talk Page. (Use four tildes to sign your post automatically with your name/timestamp.) Diverging opinions welcome! Note: a lot of good points (diverging opinions!) are covered in the mailing lists, but if we include all these various cases, split* will have 9 variants! I'm working on trying to organize all this into something meaningful.

Summary

Hacking up your own custom split (or a tokens/splitOnGlue) must be one of the most common questions from beginners on the IRC channel.

Anyone remember what the result of the "let's get split into the base library" movement's work was?

ISTR there wasn't a consensus, so nothing happened. Which is silly, really - I agree we should definitely have a Data.List.split.

A thread July 2006

http://www.haskell.org/pipermail/haskell-cafe/2006-July/thread.html#16559

A thread July 2004

http://www.haskell.org/pipermail/libraries/2004-July/thread.html#2342

Goal

The goal is to reach some kind of reasonable consensus, specifically on naming and semantics. Even if we need pairs of functions to satisfy various usage and algebraic needs. Failing to accommodate every possible use of these functions should not be a sufficient reason to abandon the whole project.

Note: I (Jared Updike) am working with the belief that efficiency should not be a valid argument to bar these otherwise universally useful functions from the libraries; regexes are overkill for 'split' and 'replace' for common simple situations. Let's assume people will know (or learn) when they need heavier machinery (regexes, FPS/ByteString) and will use it when efficiency is important. We can try to facilitate this by reusing any names from FastPackedString and/or ByteString, etc.

The Data.List functions

split (working name)

We need a few of these:

split :: Eq a => a -> [a] -> [[a]]
splitWith :: (a -> Bool) -> [a] -> [[a]]
tokens :: (a -> Bool) -> [a] -> [[a]]

That preserve:

join sep . split sep = id

See below for 'join'

And some that use above split but filter to remove empty elements (but do not preserve above property). Easy enough:

split' :: Eq a => a -> [a] -> [[a]]
splitWith' :: (a -> Bool) -> [a] -> [[a]]
tokens' :: (a -> Bool) -> [a] -> [[a]]

i.e.

split' sep = filter (not . null) . split sep

Usage would be:

tokensws = tokens' (`elem` " \f\v\t\n\r\b")

tokensws "Hello  there\n \n   Haskellers! " ==
   ["Hello", "there", "Haskellers!"]

Would a nonnulls = filter (not . null) function be a better alternative to defining separate filtered splits? Nmessenger 11:09, 28 December 2006 (UTC)

TODO: add version like python with multi-element separator

TODO: give code, copy-paste from threads mentioned above

TODO: list names and reasons for/against

replace (working name)

replace :: [a] -> [a] -> [a] -> [a]

like Python replace:

replace "the" "a" "the quick brown fox jumped over the lazy black dog"
===>
"a quick brown fox jumped over a lazy black dog"

TODO: give code, copy-paste from threads mentioned above

TODO: list names and reasons for/against

join (working name)

join :: [a] -> [[a]] -> [a]

join sep = concat . intersperse sep

TODO: copy-paste things from threads mentioned above

TODO: list names and reasons for/against

Filter

Am I the only person who thinks 'filter' is a misleading name? For example:

filter odd

Now, to me, that looks like it ought to filter out all odd values, leaving only even ones. In fact (as you all know) it does precisely the opposite.

I would suggest that 'select' would be an infinitely better name. (It works for SQL!)

MathematicalOrchid 16:56, 18 January 2007 (UTC)

Function Behavior Changes

It is dangerous to change the behavior of prelude functions, however currently unlines will add an additional, unnecessary newline. unlines . lines is effectively (++"\n") rather than id.

I propose this definition instead:

unlines = concat . intersperse "\n"

Also, lines/words will be effectively defined in terms of split.