From: Daryl Richter <daryl@...>
Date: 2005-08-30T05:21:07+09:00
Subject: Re: Method behaves differently when called using #send

Logan Capaldo wrote:
> 
> On Aug 29, 2005, at 3:54 PM, Daryl Richter wrote:
> 
>> Yukihiro Matsumoto wrote:
>>
>>> Hi,
>>> In message "Re: Method behaves differently when called using #send"
>>>     on Mon, 29 Aug 2005 08:07:55 +0900, "David A. Black"  
>>> <dblack@wobblini.net> writes:
>>> |I don't think it's an overloading, just another (i.e., other than
>>> |receiver-changing) nuance of the "more dangerous" concept.  I  actually
>>> |didn't understand Matz's concern fully.  I assume it's not
>>> |specifically about the fact that send! is not destructive, since
>>> |that's OK in a bang method.
>>> The point is not being destructive or not, since it's OK in a bang
>>> method as you described (by quoting me).  The point is that I don't
>>> really feel it's dangerous to call private methods by the new method
>>> (#fcall, or send!).  They are just different.  I'm hesitating put a
>>> bang sign to something I don't feel dangerous.
>>>                             matz.
>>>
>>
>> I'm a Smalltalker, so I agree that characterizing this as  "dangerous" 
>> is overstated.  I think "advanced feature" is perhaps  more accurate?
>>
>> I like #send the way it is.  It's an "advanced" feature and you  
>> should only use it if you understand the potential side effects.
>>
>> -- 
>> Daryl Richter
>> Platform Author & Director of Technology
>>
>> ((         Brandywine Asset Management          )
>>  ( "Expanding the Science of Global Investing"  )
>>  (          http://www.brandywine.com           ))
>>
>>
>>
> 
> It's not  a potential side-effect, its a bug that doesn't let you use  
> it the way you should be able to.  #send is a way to determine at  
> runtime what message is sent to an object, if it has a different  method 
> look-up order than a send message literal, that's "wrong". It  pretty 
> much needs to be changed. The alternate change is to keep send  the way 
> it is and have #fcall or whatever its called use the standard  lookup 
> order. Either way there needs to be a way to programmatically  send a 
> message whose semantics are the same as a message send literal.
> 

Sorry, my misunderatanding.  I thought it had to do with the fact that 
some people didn't like the fact that #send "sees" private methods and 
wanted a new explicit method for indicating that you were "breaking 
through to the private methods."

My (many) years of experience have shown me that "private" methods are 
more of a hindrance than a help.

My apologies again.

-- 
Daryl