From: "Chr. Rippel" <chr_news@...>
Date: 2002-01-19T06:59:11+09:00
Subject: Re: dynamic method creation

"Albert L. Wagner" <alwagner@uark.edu> wrote in,
....
> > >class Record
> > >       def initialize(name)
> > >               @name = name
> > >       end
> > >end
> > >
> > >fieldNames = ['abc', 'dof', 'ghi']
> > >fieldTypes = ['String', 'Fixnum', 'Float']
> > >rec = Record.new('aRecord')
> > >class << rec
> > >end
> >  ^^^
> > This 'end' is probably not supposed to be here, right?
>
> I think it's needed to close the class << rec, just above it.  I
> get a parse error without it.
wierd it should not be necessary - anyway modulo the `case issue'
you might be looking at something like this

------
class Record
  def initialize(name)
    @name = name
  end
  def singleton_attr_writer (*names)
      singleton_type = class << self; self end
      # over come the private declaration of attr_writer
      singleton_type.send (:attr_writer,*names)
  end
end

rec = Record.new  'aRecord'
fieldNames = ['abc', 'dof', 'ghi']
rec.singleton_attr_writer (*fieldNames)

p rec.instance_variables
rec.abc= 'a'
rec.dof= 1
rec.ghi= 1.0
p rec.instance_variables
inst_vars = rec.instance_variables.collect do |var|
  rec.instance_eval (var)
end
p inst_vars
-----
["@name"]
["@dof", "@abc", "@name", "@ghi"]
[1, "a", "aRecord", 1.0]
-----

/Christoph

>
> >
> > >fieldNames.size.times do | i |
> > >       f = fieldNames[i]
> > >       defStr = "def rec.#{f}=(obj)\n"
> > >       case fieldTypes[i]
> > >               when 'String'           then defStr << "@#{f} = obj\n"
> > >               when 'Fixnum'           then defStr << "@#{f} = obj\n"
> > >               when 'Float'            then defStr << "@#{f} = obj\n"
> > >       end
> > What is the purpose of this 'case' statement?  All branches seem to do the
> > same thing - am I missing something?
>
> No, you aren't missing anything.  Those are fossils from earlier
> code. Sorry.
>
> >
> > >       defStr << "end\n"
> > >       puts defStr
> > >       eval defStr
> > Maybe it's better to use instance_eval?
>
> Like this?
>
> rec.instance_eval defStr
>
> That works too.  What is the difference?
> >
> > >end
> >  ^^^
> > I suspect that this 'end' is the one that corresponds to the
> > class << rec above (?)
> This one closes: fieldNames.size.times do | i |
> >
> > >p rec.abc=(nil)
> > >p rec.dof=(nil)
> > >p rec.ghi=(nil)
> > >p rec.methods
> >
> > I can't come up with another way that doesn't use some kind of eval to do
> > this.
> >
> > Phil
>
> Thanks, Phil. I think I'll try writing it to a file and then
> require it. It shouldn't happen too often.
>
> --
> "It requires wisdom to understand wisdom:
>  the music is nothing if the audience is deaf."
>  - Walter Lippman
>