Ruby/Tk: How to access surrounding class from Tk Callback?

[Ronald Fischer]

Assume that I'm modelling a visible form, consisting of entry fields,
buttons etc., as a class:

require 'tk'
class Form
def initialize(...)
@foo=1234
...
@root=TkRoot.new() { title "My Form" }
action_button=TkButton.new(@root) {
text "Action!"
command {
# Do something with @root and @foo
puts @root.foo # DOES NOT WORK
}
}
action_button.pack("side" => "right");
end
end

The problem is with the command routine inside the action_button:
Both @root and @foo are undefined here. Obviously, the class context
is not present anymore inside this callback function.

How would one model this situation in Ruby? In C++ terms: How do
I pass the 'this' pointer of my class into the command callback
of the button?

In my case, I happen to have only one instance of the Form, so
I could use a global variable, $form, to hold a reference to my
Form object and access it via this global, but of course this
solution would be ugly beyond imagination...

Kind regards,

Ronald

 

[Hidetoshi NAGAI]

From: (e-mail address removed)-mobilphone.net (Ronald Fischer)
Subject: Ruby/Tk: How to access surrounding class from Tk Callback?
Date: Mon, 27 Feb 2006 21:18:35 +0900

@root=TkRoot.new() { title "My Form" }
action_button=TkButton.new(@root) {
text "Action!"
command {
# Do something with @root and @foo
puts @root.foo # DOES NOT WORK
}
}

It may be a FAQ.
A block given to <TkWidget-class>.new method is
evaluated by <the widget>.instance_eval(<the block>).
That is, in the block, 'self' is the widget object.
So, @root in your button's command is an instance variable
of the button widget.
To avoid it, you have to give attention to scope of variables.
For example,
------< example 1 >------------------------------------------
def initialize(...)
foo = @foo=1234
...
root = @root = TkRoot.new() { title "My Form" }
action_button=TkButton.new(@root) {
text "Action!"
command {
# use local variables
puts foo # same as @foo
puts root # same as @root
}
}
action_button.pack("side" => "right");
end
-------------------------------------------------------------
------< example 2 >------------------------------------------
def initialize(...)
@foo=1234
...
@root = TkRoot.new() { title "My Form" }
action_button=TkButton.new(@root,
:text=>"Action!",
:command=>proc{
puts @foo
puts @root
})
action_button.pack("side" => "right");
end
-------------------------------------------------------------
------< example 3 >------------------------------------------
def initialize(...)
@foo=1234
...
@root = TkRoot.new() { title "My Form" }
cmd = proc{
puts @foo
puts @root
}
action_button=TkButton.new(@root) {
text "Action!"
command cmd
}
action_button.pack("side" => "right");
end

 

[Ronald Fischer]

From: (e-mail address removed)-mobilphone.net (Ronald Fischer)

It may be a FAQ.

Arigatou gozaimasu! But where can I find the FAQ? I was able to
locate one only in Japanese, and my Japanese is barely sufficient
to order some Yakitori, let alone reading such a FAQ.... :-(

Aside from a FAQ, a Ruby/Tk reference manual might be handy. Right
now I am "learning" Ruby/Tk by looking at examples I find on the
net, and *guessing* how they could work - not a very satisfying
means to grasp a new language.

A block given to <TkWidget-class>.new method is
evaluated by <the widget>.instance_eval(<the block>).
That is, in the block, 'self' is the widget object.
So, @root in your button's command is an instance variable
of the button widget.

I understand.

To avoid it, you have to give attention to scope of variables.
For example,
------< example 1 >------------------------------------------
def initialize(...)
foo = @foo=1234
...
root = @root = TkRoot.new() { title "My Form" }
action_button=TkButton.new(@root) {
text "Action!"
command {
# use local variables
puts foo # same as @foo
puts root # same as @root
}
}
action_button.pack("side" => "right");
end

I amazed that this works, but I don't know how. foo is a local
variable inside initialize, isn't it? So it should go out of scope
as soon as initialize ends. How then can I refer to it from inside
my action command?

------< example 2 >------------------------------------------
def initialize(...)
@foo=1234
...
@root = TkRoot.new() { title "My Form" }
action_button=TkButton.new(@root,
:text=>"Action!",
:command=>proc{
puts @foo
puts @root
})
action_button.pack("side" => "right");
end

Here I have two questions: What is the meaning of the
colon in front of "command", and why can I access @root
now from inside command, but not in my original code?

------< example 3 >------------------------------------------
def initialize(...)
@foo=1234
...
@root = TkRoot.new() { title "My Form" }
cmd = proc{
puts @foo
puts @root
}
action_button=TkButton.new(@root) {
text "Action!"
command cmd
}
action_button.pack("side" => "right");
end

This looks clever too. From a design point of view, I like solution 2
best, because it avoids the need for additional variables. Still would
like to understand, *why* it works.

Ronald

 

[Ronald Fischer]

Here I have two questions: What is the meaning of the
colon in front of "command", and why can I access @root
now from inside command, but not in my original code?

I think, I figured this out by myself: Instead of setting
text and command in an initialization procedure for the button,
you supply them as parameter command denotes the named parameter
"command" - I had used only positional parameters before, so I
was not aware of this possibility). Since the command procedure
is defined inside the list of actual parameters to the TkButton
constructor, they can access any variable within the scope of
the initialize function body; in particular, they can access
the member variables of the surrounding class.

Ronald

 

[Hidetoshi NAGAI]

From: (e-mail address removed)-mobilphone.net (Ronald Fischer)
Subject: Re: Ruby/Tk: How to access surrounding class from Tk Callback?
Date: Wed, 1 Mar 2006 17:48:37 +0900

Arigatou gozaimasu! But where can I find the FAQ?

If I am right, I'd wrote a same answer on this ML. ;-)

Aside from a FAQ, a Ruby/Tk reference manual might be handy. Right
now I am "learning" Ruby/Tk by looking at examples I find on the
net, and *guessing* how they could work - not a very satisfying
means to grasp a new language.

I want to compose Ruby/Tk manuals/documents.
But I don't have enough time and collaborators to do that.
And it is so hard for me to translate them to English,
because I'm poor at English.

(snip)
I amazed that this works, but I don't know how. foo is a local
variable inside initialize, isn't it? So it should go out of scope
as soon as initialize ends. How then can I refer to it from inside
my action command?

A Proc object keeps its binding.

(snip)
Here I have two questions: What is the meaning of the
colon in front of "command",

":command" is a symbol.
Current Ruby/Tk can accept a string or a symbol as a key of an option.

FYI, a has of options is a little faster than method calls.
For example,
---------------------------------------------------------------------
action_button=TkButton.new(@root, :text=>"Action!",:command=>cmd)
---------------------------------------------------------------------
this calls the Tk interpreter only once.
But,
---------------------------------------------------------------------
action_button=TkButton.new(@root) { # <= 1
text "Action!" # <= 2
command cmd # <= 3
})
---------------------------------------------------------------------
this calls Tk interpreter three times.

and why can I access @root
now from inside command, but not in my original code?

It depends on the difference of scope.
In the block given to TkButton.new, 'self' is the button object.
But the proc object is defined in the scope of 'initialize'.

(snip)
This looks clever too. From a design point of view, I like solution 2
best, because it avoids the need for additional variables. Still would
like to understand, *why* it works.

The answer is same as the one for < example 1 >.