Re: [cedet-eieio] eieio.el: another constructor question

[Eric M. L. <er...@si...>]

Hi Jan,

Thanks for the detailed report, analysis, example, and even a patch! 
Huzzah!

As for your final question on if this is better, I found the following 
doc on clos.org:

---
When call-next-method is called with no arguments, it passes the current 
method's original arguments to the next method.
---

which indicates to me that you have provided the right solution.

It even still passes my unit tests.

Thanks!
Eric

Jan Moringen wrote:
> Hi,
> 
> while working with a class hierarchy similar to this
> 
>  a
> / \
> b c
> \ /
>  d
> 
> I discovered surprising (to me) behavior. Each class has an
> initialize-instance method and they run in one of the expected orders:
> d, b, c, a.
> 
> The surprising part: each initialize-instance method uses
> call-next-method. The methods of d and b provide replacement arguments
> and extent the list of initargs with :c 'c and :d 'd respectively.
> However, when c uses (call-next-method) the eieio-generic-call-arglst of
> the original call (to initialize-instance of d) is used. I expected the
> replacement arguments provided in the initialize-instance methods of d
> and b to be propagated to the call of c's initialize-instance.
> 
> With instrumented methods that print their arguments, the following
> happens:
> 
> (d "" :a 'a)
> | d (:a a)
> | b (:a a :d d)
> | c (:a a :d d :b b)
> | a (:a a)           ; (call-next-method) in c calls a's 
>                      ; initialize-instance with original argument list
> 
> I expected the result to be more like this:
> 
> (d "" :a 'a)
> | d (:a a)
> | b (:a a :d d)
> | c (:a a :d d :b b)
> | a (:a a :d d :b b) ; initialize-instance of a is called with modified 
>                      ; argument list
> 
> I produced the second trace by changing the following in eieio.el
> (call-next-method):
> 
> @@ -2174,6 +2222,8 @@
>  	(apply 'no-next-method (car newargs) (cdr newargs))
>        (let* ((eieio-generic-call-next-method-list
>  	      (cdr eieio-generic-call-next-method-list))
> +	     (eieio-generic-call-arglst
> +	      newargs)
>  	     (scoped-class (cdr next))
>  	     (fcn (car next))
>  	     )
> 
> Is this an improvement?
> 
> Kind regards,
> Jan
> 
> 
> ------------------------------------------------------------------------
> 
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
> trial. Simplify your report design, integration and deployment - and focus on 
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> cedet-eieio mailing list
> ced...@li...
> https://lists.sourceforge.net/lists/listinfo/cedet-eieio