Juha - potential bugs

[squirest]

Hey,

It seems you're actively working in the code right now... I think you might run into problems because of the following bugs.

ModelMBeanInfoSupport constructor which uses the *Info[] arrays it's passed *must* check for null arrays.

If it doesn't then you may see classcast exceptions when doing something like this:

ModelMBeanConstructorInfo[] foo = (ModelMBeanConstructorInfo[]) infoSupport.getConstructors();

It's because the super() constructor that takes those arrays will create MBean*Info[] arrays rather than ModelMBean*Info[] arrays.

Also the ModelMBean*Info classes are not creating default descriptors. This will cause code to blow up with NPEs all over the place.

Trev

[squirest]

> ModelMBeanInfoSupport constructor which uses the
> *Info[] arrays it's passed *must* check for null
> arrays.

Fixed.

[juha]

you're right, looking at it now.... there's also some weird things going on with clone...

debug time

[juha]

I've got the default descriptor for attributes done...
debugging clones now

[juha]

right, MMB attribute and operation info should behave better now... if you wanna clean up the rest of the *info classes, take it away..

-- Juha

[squirest]

MMBConstructorInfo looks ok which leaves MMBNotificationInfo - I'll stick that in now.

I'll also implement the copy constructors - strangely MMBConstructorInfo doesn't have one...

After that, I guess it's human readable toString()s according to fixmes - is that in the spec?

Trevor