From: matz@... (Yukihiro Matsumoto)
Date: 2002-03-15T23:36:44+09:00
Subject: Re: key comparision inside Hash

Hi,

In message "Re: key comparision inside Hash"
    on 02/03/15, Vladimir Dozen <Vladimir.Dozen@osw.com.ru> writes:

|> Because the default semantic of "eql?" is equivalence ("equivalence"
|> means true iff two objects are same object).
|
|   Mmm. I guess you mean "equivalent", not "same":

I meant "same".  Two objects need to be same for "eql?" to be true.

|> One possible new semantic is that if object is T_OBJECT then "hash"
|> and "eql?"  test would be based on its class and instance variable
|> values.  RCR?
|
|   Oops, I fear, I cannot understand it. You express semantics in terms
|   of realization, right? ;)

Ah, yes.  It's the easiest language for me to express the behavior.

|   All I want is that default Object::hash() method return same value for
|   two objects having same number and names of object member variables,
|   and same values of those member variables. For simple cases like above
|   Coord class (where only @x and @y present) it would be enough, for
|   complex classes programmer will have to write own hash().
|
|   I guess, it will not break things?

That's what I meant in above statement.  I can't foresee any problem,
but it doesn't mean there will be no problem.

							matz.