Re: First-class data types

In article 92-03-024 lotus!wildbill@uunet.uu.net (Bill Torcaso) writes:
          There is a nit to pick with the definition of 'first class datatype'
      that asserts any operation can be applied to any object of any first class
      datatype. Consider a language in which functions are first-class:


          What is the XOR of two functions? What is the AND of two functions?


The term `first class' is used in conjunction with Scheme to mean
that you can
    - store a first class value in a data structure
    - pass it as a parameter
    - perform to it any operations *that are defined for it*
    - some other things I may have missed


I would say that prohibiting passing a function as an argument
definitely makes it second class, whereas not defining some
operations, such as XOR, for it does not affect its fundamental
status. I do not view passing a value as a parameter as applying an
operation to the value, but as a entirely different thing.


There is a difference in a value being first class and a type
being first class. A first class datatype would mean by the
above definition that you can, for instance, pass a type as a
parameter. I tend to think that a `first class datatype' means a
type whose instances have a first class status, though.
--
Pertti Kellom\"aki (TeX format)
    Tampere Univ. of TeXnology
            Software Systems Lab
[Similar comments received from Geoffrey Clem and Markus Freericks. -John]
--

Return to the comp.compilers page.
Search the comp.compilers archives again.