PR# 3426 VTCG-6 not checked

It seems to me that VTCG-6, described in ETL3 4.82-00-00
page 262 (chapter 12) is not enforced by ISE Eiffel. This
rule says that the actual generic parameters should have as
creation procedures the procedures listed in the Constraint_creators
clause of the constraint associated with their corresponding
formal parameters. So in the example below:

------------------------
class AA

creation

	make

feature

	make is
		do
			create foo
			foo.f
		end

	foo: FOO [COMPARABLE]

end
------------------------
class FOO [G -> COMPARABLE create default_create end]

feature

	f is
		local
			a: G
		do
			create a
			print (a.generating_type)
		end

end
------------------------

the compiler should have complained because in class
A `foo' is declared of type FOO[COMPARABLE], but 
`default_create' is not a creation procedure of the
actual generic parameter COMPARABLE (indeed, this
class is deferred).

Instead of that, the program crashed at runtime when
attempting to create an instance of the deferred class
COMPARABLE.

Note that I didn't check yet what effect fixing this
bug would have on AXA's source code. I will implement
the checking of VTCG-6 in my Gobo lint tool and run
it on AXA's source code to see if any of our classes
violates this rule. I'll let you know the result.


Also, I noticed that ISE Eiffel reports a syntax error
when compiling:

class A [G -> B create end]
end

But, although not very useful, ETL3 seems to allow empty
Constraint_creators clauses. Indeed, Feature_list used
in ETL3 4.82-00-00 page 261 (chapter 12) is declared in
page 136 (chapter 7) as a list of zero or more Feature_name.

From: Eric Bezault <ebezault@axarosenberg.com>
To: "'bugs@berkeley.eiffel.com'" <bugs@berkeley.eiffel.com>
Cc: "'manus@eiffel.com'" <manus@eiffel.com>,
        RC_Tools
	 <RC_Tools@axarosenberg.com>
Subject: RE: Compiler/3426
Date: Sun, 18 May 2003 18:29:09 -0700

 I guess I got too much sun on my head this afternoon.
 The rule I was talking about is not VTCG-6 but VTCG-4.
 Furthermore it seems that this rule is actually enforced
 when the actual generic parameter is not deferred:
 
 ------------------------------
 class AA
 
 creation
 
 	make
 
 feature
 
 	make is
 		do
 			create foo
 			foo.f
 		end
 
 	foo: FOO [STRING]
 
 end
 ------------------------------
 
 although the error message is quite confusing:
 
 ------------------------------
 Error code: VTCG
 Error: actual generic parameter does not conform to constraint.
 What to do: make sure that actual parameter is a type conforming to the
   constraint (the type appearing after `->' for the corresponding formal).
 
 Class: AA
 In declaration: FOO [
....
Output truncated, Click download to get the full message