--- "Miller, Raul D" <[EMAIL PROTECTED]> wrote:

> I don't see any advantage of the name pbx_create over 
> create_pbx_ when they both refer to the same thing, 
> and create_pbx_ is simpler to understand.

That's the whole idea: to prevent locale switch.
Whenever you say verb_loc_ you are no longer in your locale.
Fix (f.) would have fixed it by instead executing a copy.
But locked script instead gives you back verb_loc_ and
you again end up in the other locale.

This double def cover thing is a brillinat idea, 
but it doesn't feel right to need to do it.


> One of these days, someone is going to figure out 
> how to reliably deploy versions of these names 
> which rely solely on the 18!: facilities and 
> entirely discard the COXXX variables.

This is interesting. So what is the relationship
between COCLASSPATH and 18!:2 ?
COCLASSPATH is a redundant convenience facility 
to store up the path for the sole purpose of 
conew to do copath on class instances.

Interestingly, why not just store the 
instance path-to-be as the copath of
the class locale, and then just copy class path
to instance path?

Instead of
  COCLASSPATH__c copath obj
have
  (copath c) copath obj


> That, I think, was the original J design.
> 
> Anyone really needing the current implementation
> can (and should) include a script giving 
> coextend_z_ (and so on) their current, overly fussy 
> definitions.


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to