MBean attributes shouldn't be tied to MBean instance variables. about jmx HOT 4 CLOSED

You comment is right in that if you call attr_accessor and then def the same methods they will override. The opposite is also true however. If you attr_accessor after you have defined your methods then attr_accessor versions of the methods will replace your existing ones.

Here is an idea. We could stop using @ivar for storage and only use gett/setter methods (and if none are defined then define them)? As it stands right now attribute behavior is a bit destructive in that it assumes it can just make getter and setter methods. I am going to ask Jay to comment since he made the original impl of attributes.

from jmx.

I changed the behavior of this in 5c0295f and hopefully all parties will be happy. There are new optional fields for the *_attribute methods. The extra arguments exist for specifying the name of the method to send to if provided:

r_attribute :explicit_reader, :int, "Sample int with writer", :my_reader
w_attribute :explicit_writer, :int, "Sample int with writer", :my_writer
rw_attribute :explicit_both, :int, "Sample int with writer", :my_read, :my_write

def my_reader
    @domain_object.real_value
end

This also satisfies a larger requirement that the Dynamic MBean does not need to be the real source of data. Define your own method which then asks the real source of data for your attribute value. Thanks for filing this. It made me step back a bit on this feature. Also let me know if this works for you and then I will push 0.6.

Jay, the only real change between what we had before and what we have now if you do not provide explicit read/write methods is that we will not override via attr_reader, etc.. if
there is already a method of the same name.

from jmx.

At first glance it looks good to me.. I'll spend more time reviewing over the next day or so... work's been crazy lately :P

from jmx.

It looks good to me too. Thanks!

from jmx.