Re: is_equal of HASHABLE

Frank Boehme wrote:

> Yes, to me that looks more suitable for general purpose programs, i.e.
> where it is the intention just to make objects hashable.
> BTW, in my G2F3 code a more special situation arises since hundreds of
> thousands of game positions may be stored in the transposition table.
> Obviuously, I am concerned about memory usage which is why I store only
> that part of information which determines whether two positions are
> identical or different and who is to move next. This is what the feature
> {STORAGE_HASHABLE}.hash_key: HASH_KEY does. Now, in that (admittedly
> special) case, HASH_KEY key inherits only from HASHABLE and its
> attributes are only expanded types so that the built-in definition of
> is_equal will always do. But I agree that Dominque's suggestion (defer
> is_equal in HASHABLE and adding the psotcondition) is a good decision:
> It guarantees that the assumptions which are made in HASHED_SET and
> HASHED_DICTIONARY are checked by a contract and it does not impose much
> hassle. 
> cheers,
> Frank
Obviously, if some class as only one parent which is HASHABLE, it is 
possible
to get the ANY body by adding an extra INSERT ANY.
I have just applied the new definition on our working version.
Here is the commit message:
-----------------------------------------------------------------
Module name:	SmartEiffel
Repository:	SmartEiffel/tools/misc/
Changes by:	colnet	05/05/17 15:28:18
CVS server:	cvs-sop.inria.fr

Modified files:
	SmartEiffel/lib/abilities/: hashable.e
	SmartEiffel/lib/kernel/: hierarchic_graph_node.e
	SmartEiffel/lib/numeric/: integer_general.e
	                          mutable_big_integer.e number.e
	                          real_general.e
	SmartEiffel/lib/string/: character.e string.e unicode_string.e
	SmartEiffel/lib/time/: microsecond_time.e time.e
	SmartEiffel/tools/: class_name.e class_text.e feature_name.e
	                    hashed_string.e live_type.e
	SmartEiffel/tools/expression/: agent_creation.e
	                               manifest_generic.e
	                               manifest_string.e
	SmartEiffel/tools/feature/: native_plug_in.e
	SmartEiffel/tools/misc/: agent_args.e graph_node.e plugin.e

Log message:
	To follow the proposal of Frank Boehme ([email protected]) and
         the
	result of the recent discussion on our mailing list about the
	redefinition of `is_equal´ in HASHABLE. The new definition now
         has
	an extra "ensure then" assertion and is redefined as deferred !
	
	Actually, the modification of HASHABLE forced me to remove many
	"redefine is_equal" here and there (which is actually a good
         news).
-----------------------------------------------------------------
Except in case of bad surprise while running all tests, I think we won't
roll back. Good.
Thanks again,
-- 
--------------------------------------------------------------
[email protected] -- IUT (Nancy 2) -- INRIA Lorraine
http://SmartEiffel.loria.fr  --  The GNU Eiffel Compiler
POST: Loria, B.P. 239,54506 Vandoeuvre les Nancy Cedex, FRANCE
Voice:+33 0383913140 Mobile: +33 0665362381 Fax:+33 0383581701