Home/ Questions/Q 5939081
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-05-22T15:46:33+00:00

I’d like to define a fixed set of values, they are all integral, but

  • 0

I’d like to define a fixed set of values, they are all integral,
but elements are with different type(Int8/Int16/Int32/Integer).

the type of each element is fixed, and there will be only get/set
functions. like ‘java beans’.

with my intuition, It may be:

data MyFixedSet = MyFixedSet {getva :: Int32, getvb :: Int8,
getvc :: Integer, … , getvxx :: Int16}

or use tuples:

data MyFixedSet = MyFixedSet (Int32,Int8,Integer, … Int16)

or :

type MyFixedSet = (Int32,Int8,Integer, … Int16)

then I what to have functions to operate with it:

getVA (va,,,…..,) = va
setVA (,vb,vc,…..vxx) va = (va,vb,vc,…vxx)
getVA (,vb,,…..,) = vb
setVA (va,,vc,…..vxx) vb = (va,vb,vc,…vxx)
…

but I think they are too heavy to use, in my case, there will be
200~500 elements in a set.

Should I use Map?
if I do this:

data Elem = X8 Int8 | X16 Int16 | … | XI Integer
let set = Map.fromList [(0, X32 1234), (1, X8 666), …]
then I have to do some Type-check when I extract things from Map.

I’d like to know is there is a ‘beautiful’ and ‘efficient’ way to do this?

how about their performance?

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    The record-based solution may indeed be too heavy, especially when updating values. Considering the values in Haskell are immutable, every record update requires construction of a new object. In your case this object will be quite large, so its construction may be slow. Maps should be faster.

    Now, to make Maps more type-safe, How about storing one map per your data type?

    data MyFixedSet = MyFixedSet (IntMap Int8) (IntMap Int16) (IntMap Int32) (IntMap Integer)

    This way you don’t have to type-check extracted values, but you have to lookup the right map.

    Returning to the record-based solution, one can improve performance by splitting the large data structure in several parts, like a tree:

    data MyFixedSet = MyFixedSet Part1 Part2 Part3 Part4
data Part1 = Part1 Part1SubPart1 Part1SubPart2 Part1SubPart3
....
data Part1SubPart1 = Part1SubPart1 Int32 Int8 Integer
....

    You can even do this automatically and generate some nice accessor functions with the help of Template Haskell.

    • 0