Hi,
I was playing with various modules and found something I can't
explain. It seems like module inclusion's order matters ? Here is a
simple test case :
module B; end
module C; include B; end
module A; def somemethod; end; end
#reopening
module B; include A; end
p B.instance_methods #=> ["somemethod"]
p C.instance_methods #=> [] ???
I don't understand why C doesn't inherit A's instance_methods. Does
somebody know something about that ?
Hello,
I was playing with various modules and found something I can't
explain. It seems like module inclusion's order matters ? Here is a
simple test case :
[Snip example]
I don't understand why C doesn't inherit A's instance_methods. Does
somebody know something about that ?
Yes, I asked a similar version of this question in [ruby-talk:121118]
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/121118
Pit Capitain provided a good summary and link to further details in
[ruby-talk:121191]
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/121191
I hope this helps,
Zev
On Tue, 20 Jun 2006 07:32:22 +0900, Jonas Pfenniger <zimba.tm@gmail.com> wrote:
This is called the "Dyanmic Module Inclusion Problem". And Pit is
right, it is an implmentation issue. I think Matz would love to have a
fix for it, but appearently it is difficult, if not impossible, to do
so efficiently,
T.
Thanks both of you for your insight. I've created a small hack to fix that issue
class Module
# Weird hack to ensure deep inclusion of modules
def include_deep(mod)
include mod
ancestors[1..-1].select do |a|
a != mod and not a.ancestors.include?(mod)
end.each { |a| a.include_deep(mod) }
end
end
Jonas Pfenniger wrote:
Thanks both of you for your insight. I've created a small hack to fix that issue
class Module
# Weird hack to ensure deep inclusion of modules
def include_deep(mod)
include mod
ancestors[1..-1].select do |a|
a != mod and not a.ancestors.include?(mod)
end.each { |a| a.include_deep(mod) }
end
end
Hmm... Doesn't seem to do anything. Notice even without the above one
can instantiate a class and the mothod will "magically" showup. From
your example:
class X
include B
end
X.new.somemethod
T.
Hmm right, let me correct that
On 20/06/06, transfire@gmail.com <transfire@gmail.com> wrote:
Hmm... Doesn't seem to do anything. Notice even without the above one
can instantiate a class and the mothod will "magically" showup. From
your example:
doh' I would need to find the children and not the ancestors [C] of
the module (B) in which in include the other (A)... I guess it's
doable by browsing the ObjectSpace but it's quite expensive. If only
modules and classes would also list their childrens. Which is also
doable by overriding Module#included I guess but it's not really
reliable isn't it ?
On 20/06/06, Jonas Pfenniger <zimba.tm@gmail.com> wrote:
On 20/06/06, transfire@gmail.com <transfire@gmail.com> wrote:
> Hmm... Doesn't seem to do anything. Notice even without the above one
> can instantiate a class and the mothod will "magically" showup. From
> your example:Hmm right, let me correct that
--
Cheers,
zimbatm