Robert Klemme wrote:
"Benny" <linux@marcrenearns.de> schrieb im Newsbeitrag
news:2npa39F2ppenU1@uni-berlin.de...
hi all,
its time for stupid questions again 
is there a way to get the name of an object, e.g.
test = MyClass.new
test.name #=> "test"
That's not the name of the object but the name of a variable that
references the object. What would you expect to be the name in this case:
test = MyClass.new
t2 = test
# name?
ok, see my other posting:
"substitute Object#name with Object#names
in my first posting"
so it would be
test = MyClass.new
t2 = test
t2.names #=> ["t2", "test"]
test.names #=> ["t2", "test"]
This isn't exactly what you want, but it's as close as you can come, currently, in ruby. Without a whole bunch of ugliness, anyway.
def names(obj, bind)
vars = eval("local_variables", bind)
vars.select{|var| eval(var, bind).id == obj.id}
end
==>nil
a = 2.3
==>2.3
b = a
==>2.3
c = 2.3
==>2.3
names(a, binding)
==>["a", "b"]
names(c, binding)
==>["c"]
names(2.3, binding)
==>
Now I suppose it would be possible to do something like this (untested):
def add_binding(bind)
$stored_bindings ||=
$stored_bindings << bind
end
def do_something
add_binding(bind)
#... do other stuff
end
then...
def find_names(obj)
variable_names =
$stored_bindings.each do |bind|
vars = eval("local_variables", bind)
variable_names << vars.select{|v| eval(v, bind).id == obj.id}
end
end
Now, you can call find_names on any object that is in a binding that you have indexed.
Still, though, for tis to be a complete solution, you woiuld have to come up with a way to index the @vars and @@vars, $globals and CONSTANT_VARS, as well. I have a feeling that all this infrastructure would cause a considerable slowdown, though.
I know we have it for classes (and yes: I could use a subclass instead)
That's a special case.
yes I know: class names are constants
Consider also:
class Foo;end
=> nil
Foo
=> Foo
f = Foo
=> Foo
f.name
=> "Foo"
f == Foo
=> true
I dont get you here 
I think as names of classes a constants and names of variables are symbols
they are a entirely different thing (in the way ruby handles it)
I only made the statement with the subclass to get no such answer from the
list ("use a subclass if you need names"). and as I didnt want to use
constants I also didnt want to use a subclass
but is there a way to do it with objects?
No.
from my other posting:
"I think in the kernel we have symbols attached to object-ids (do we?)
so why not have a method to show us the symbols to a corresponding
object_id?"
If I understand correctly, they are stored that way in the internal symbol table. That symbol table is inaccessible, though. Once you are in the runtime, the name -> obj correlations are spread out in various constructs. Some are easier to get at than others; it's easy to lookup a constant, but local variables are truely a pain.
btw. why didn't matz make Class and Object be the same thing?
Err... Because they are not the same concept.
if they behave mostly the same way in ruby
it might be reflected behind the scenes.
are the differences in the abilities of Object and
Class so big that they legitimate hard distinction in the
interpreter? or are we loosing something if Object and Class
is based on the same fundamentals apart from syntactical restrictions
and the separated name spaces (constants vs. symbols)?
Class is a template for creating objects. Object is a subclass of class. This is a fairly common OO topic, and there is a pretty good description of classes and objects here:
···
On Aug 9, 2004, at 7:21 AM, Benny wrote:
Object is an instance of
Class (as Class is also, which sometimes confuses people). Object defines
everthing an instance is capable of (ok, together with Kernel, but that
contains mostly methods that one uses as functions). Could be that the
self referenciality confused you - sometimes it's hard to grasp.
I think its not confusing me on the contrary: I even vote for narrowing the
differences.
Maybe you should get yourself a book about programming languages concepts
and / or about OO in special. Just an idea.
I read the whole pragmatic programmer (1st ver.) and "Programmieren mit
Ruby" /Röhrl/Schmiedl/Wyss several times and I thought it would be
sufficient. but perhaps you have a good recommendation for a pure OO book.
kind regards,
benny
Anyway, that was mainly a statement to introduce the link, which does a much better job of stating things than I did...