Re: [Sbcl-devel] Bug in the CLOS MOP...

[Pascal C. <pc...@p-...>]

On 28 Jan 2008, at 18:18, Christophe Rhodes wrote:

> Pascal Costanza <pc...@p-...> writes:
>
>> Hm, has this bug report been noted? I want to make sure that this
>> doesn't go unnoticed, so a quick ack would be nice.
>
> I've tagged it, but on the other hand I have tagged about 900 messages
> in sbcl-devel, some dating from 2003 or so, so...

Thanks a lot. That's already good to know... ;)

> (rapid question: could you simply be missing a finalize-inheritance
> somewhere?)

It's true that at least one of the involved classes isn't finalized  
yet, BUT: You don't need finalized classes / class precedence lists  
for implementing subtypep ("subclassp")!!! Section 4.3.1 doesn't need  
the class precedence list to define the term "subclass", and Section  
4.3.7 refers to that definition of "subclass" to define the semantics  
of subtypep. Furthermore, the CLOS MOP requires that compute-class- 
precedence-list returns a list that contains the respective class and  
_all_ superclasses, so finalize-inheritance / compute-class-precedence- 
list cannot violate the subclass relationship. ("Otherwise the results  
are undefined.")


Pascal

-- 
1st European Lisp Symposium (ELS'08)
http://prog.vub.ac.be/~pcostanza/els08/

Pascal Costanza, mailto:pc...@p-..., http://p-cos.net
Vrije Universiteit Brussel, Programming Technology Lab
Pleinlaan 2, B-1050 Brussel, Belgium