Re: is_equal of HASHABLE

Le vendredi 13 Mai 2005 13:51, Frank Boehme a écrit :
> Frank Boehme wrote:
> > > Doing such thing would break codes that use SET to record objects by
> > > there address (hash_code: INTEGER is do Result :=
> > > to_pointer.hash_code). Maybe such implementation may be usefull.
>
> More about that: I think such code should only be used for hashing if
> only the "fast-.." features are used for lookup, i.e. the ones based on
> '=' rahter than 'is_equal'.
>
> F

yes indeed

i was wrong.

for the kind of need about wich i wrote, a class as above is enougth.

expanded class ?????[X] 
inherit HASHABLE COMPARABLE
feature
	item : X
	set_item(x: like item) is do item := x end
	hash_code: ITENGER is do item.to_pointer.hash_code end
	infix "<"(other: like Current): BOOLEAN is do 
		Result := pointer_to_integer(item.to_pointer)
			< pointer_to_integer(item.to_pointer) end
feature {}
	pointer_to_integer(p: POINTER): INTEGER_64 is external
		"C inline" alias "(int_64_t)($p)" end
end

But i can not find a name for such kind of utility class (IDENTIFICATION?, 
ALIAS?, ISOLATOR?)

regards

jb