Const_defined? behavior


#1

Is this the intended behavior of const_defined?

IO::SYNC
=> 4096

IO.const_defined? :SYNC
=> false

It appears that const_defined? only returns true if the class is the
originator of the constant. If the constant was borrowed from
somewhere else (in this case, File::Constants), it returns false.

That’s on purpose, right? If that’s the case, I must resort to
IO.constants.member?(‘SYNC’) for my intended behavior, I suppose.
Seems a little strange.

Dan Amelang


#2

Hi

At Mon, 16 Jan 2006 17:05:29 +0900,
Daniel A. wrote in [ruby-talk:175899]:

Is this the intended behavior of const_defined?

IO::SYNC
=> 4096

IO.const_defined? :SYNC
=> false

It appears that const_defined? only returns true if the class is the
originator of the constant. If the constant was borrowed from
somewhere else (in this case, File::Constants), it returns false.

Yes.

That’s on purpose, right? If that’s the case, I must resort to
IO.constants.member?(‘SYNC’) for my intended behavior, I suppose.
Seems a little strange.

defined?(IO::SYNC)


#3

Hi Nobu,

defined?(IO::SYNC)

In my case, the constant’s name is stored in a variable. So, I’d have
to resort to doing an ‘eval’ if I were to follow your suggestion. So
at least I have two options now:

Option 1 (Nobu’s suggestion)

found = eval(“defined?(obj::#{const})”)

Option 2 (My original idea)

found = obj.respond_to?(:constants) && obj.constants.member?(const)

Any better ideas? I usually try to stay clear of ‘eval’ if I can.

Dan Amelang


#4

Pretty cool!

Not what I was looking for, but may prove useful later.

Dan


#5

On Jan 16, 2006, at 5:52 PM, Daniel A. wrote:

Option 2 (My original idea)

found = obj.respond_to?(:constants) && obj.constants.member?(const)

Any better ideas? I usually try to stay clear of ‘eval’ if I can.

Dan Amelang

def const_defed(var)
var.split(/::/).inject(Object) do |left, right|
begin
left.const_get(right)
rescue NameError
break nil
end
end
end