|
|
last edited 16 years ago by gdr |
1 2 3 4 5 6 7 8 9 10 11 | ||
Editor: gdr
Time: 2008/05/29 17:15:33 GMT-7 |
||
Note: |
added:
From gdr Thu May 29 17:15:33 -0700 2008
From: gdr
Date: Thu, 29 May 2008 17:15:33 -0700
Subject:
Message-ID: <20080529171533-0700@axiom-wiki.newsynthesis.org>
Status: open => fixed somewhere
The code generation failure is fixed. So, I don't
think is an issue any more.
In OpenAxiom? one should be able to write a function that returns a 'Category':
axiomST(x:Integer):Category == (x=0 => SetCategory;Type) Function declaration ST : Integer -> Category has been added to workspace.
axiomST(0) Internal Error Interpreter code generation failed for expression (IF (= |#1| 0) |SetCategory| |Type|)
but we get the error message:
Internal Error Interpreter code generation failed for expression (IF (= |#1| 0) |SetCategory| |Type|)
This is ok for functions that return 'Domain':
axiomFI(x:Integer):Domain == (x=0 => Float;Integer) Function declaration FI : Integer -> Domain has been added to workspace.
axiomFI(0)
Compiling function FI with type Integer -> Domain ; (DEFUN |*1;FI;1;initial| ...) is being compiled. ;; The variable |*1;FI;1;initial;MV| is undefined. ;; The compiler will assume this variable is a global.
![]() | (1) |
axiomFI(1)
![]() | (2) |
However, in the interpreter we should probably try to accomodate for as much as we can because we can also do a small step semantics, e.g. interpret literally the codes.
Going back to your problem, an issue here is with what the type of Type should be. Currently, OpenAxiom? says that Type has type Type, instead of Category:
axiomType
![]() | (3) |
axiomtypeOf Type
![]() | (4) |
Another option might be to say that Type has type Category. That would make your function OK. Note that the following works fine
axiomST2(x: Integer): Category == (x=0 => SetCategory; Ring) Function declaration ST2 : Integer -> Category has been added to workspace.
axiomST2 0
Compiling function ST2 with type Integer -> Category ; (DEFUN |*1;ST2;1;initial| ...) is being compiled. ;; The variable |*1;ST2;1;initial;MV| is undefined. ;; The compiler will assume this variable is a global.
![]() | (5) |
axiomST2 1
![]() | (6) |
(1) -> ST(x:Integer):Category == (x=0 => SetCategory;Type) Function declaration ST : Integer -> Category has been added to workspace. Type: Void (2) -> A:=ST(0) Compiling function ST with type Integer -> Category ;;; *** |*1;ST;1;frame1392| REDEFINED ; (DEFUN |*1;ST;1;frame1392| ...) is being compiled. ;; The variable |*1;ST;1;frame1392;MV| is undefined. ;; The compiler will assume this variable is a global. (2) SetCategory Type: Category (3) -> B:=ST(1) (3) Type Type: Category (4) -> C:Category Category is a category, not a domain, and declarations require domains. (4) -> C:Category:=Ring Category is a category, not a domain, and declarations require domains. (4) -> C:=Ring (4) Ring Type: Category (5) -> )display type C Type of value of C: Category (5) -> X:Domain:=Ring -- Ring is not a domain <a href="#eq5">(5)</a> Ring Type: Category
-------
If the type of a category-valued variable can be implicitly
made to be Category
, it should also be possible to declare
this to be so.
The code generation failure is fixed. So, I don't think is an issue any more.