From: "Guy N. Hurst" <gnhurst@...>
Date: 2002-04-21T13:38:44+09:00
Subject: Re: problem with alias

"Hal E. Fulton" wrote:
> ...
> > In short,
> >
> > alias old_meth meth
> > alias old_meth meth
> >
> > will result in meth being lost forever.
> 
> [snip example]
> 
> That's interesting. I never thought of that.
> 
> A workaround (if you see this happening)
> would be to check whether the target is
> defined first...
> 
>    if !defined? meth then alias oldmeth meth end
> 
> Clunky, yes.
> 
> And don't be tempted to use the modifier form:
> 
>    alias oldmeth meth if !defined? oldmeth
> 
> It won't work. Think why.  :)
> 

Did you try both these workarounds? I already tried
these and others, to no avail. It still performed
the alias as if no check was done.

irb(main):114:0> class String
irb(main):115:1>   if !defined? oldsize then alias oldsize size end
irb(main):116:1>   def size
irb(main):117:2>     print "new version:"
irb(main):118:2>     oldsize
irb(main):119:2>   end
irb(main):120:1> end
nil
irb(main):121:0> "".size
new version:new version:0
irb(main):122:0> class String
irb(main):123:1>   if !defined? oldsize then alias oldsize size end
new version:new version:new version:irb(main):124:1>   def size
new version:irb(main):125:2>     print "new version:"
new version:irb(main):126:2>     oldsize
new version:irb(main):127:2>   end
irb(main):128:1> end
nil
irb(main):129:0> "".size
new version:new version:new version:new version:new version:new version:new version:new version:new version:new
version:new version:new version:new version:new version:new version:new version:new version:new version:new version:new
version:new version:new version:new version:new version:new version:new version:new version:new version:new version:new
version:new version:new version:new version:new version:new version:new version:new version:new version:new version:new
version:new version:new version:new version:new version:new version:new version:new version:new version:new version:new
version:new version:new version:new version:new version:new version:new version:new version:new version:new version:new
version:new version:new version:new version:new version:new version:new version:new version:ne

> As for the send_original idea, I think that's not such
> a bad idea... would it be practical to make this one
> method "fixed in place" (can't be aliased, redefined,
> or removed?). That would fix the infinite regression
> problem and would not necessitate a new keyword.
> 

But maybe to make it fixed will require a new keyword?

> Just a thought. No doubt Matz has a better idea.
> 

I hope so.


Guy N. Hurst