Re: First-class data types

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?


> What is the function-invocation of the integer constant 17? Of the
> floating-point constant 0.5?


These operations are well defined--they all generate errors.


Most people mean 1st class data types have the same language *rights*.


    Does it need to have a name?


    Can it be created and returned from within a function, passed as a
    parameter, stored in a data structure?


More generally, do all data objects satisfy basic design principles
using the same rules? [E.g. see R. Tennent: _Principles of
Programming Langauages_, Prentice Hall, 1981, ISBN 0-13-709873-1
    - Principle of Abstraction
    - Principle of Correspondence
    - Principle of Qualification
].


For example, in the C language function *pointers* are first class,
but functions are not--you cannot create an unnamed function.




-Ken Dickey kend@data.rain.com
--

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