Am 07.10.2005 um 20:13 schrieb Ken Causey:

> I was helping someone today with IRCe and happened to try to use it  
> in a
> 3.9-6693md4 image.  I got a debugger immediately and tracked the  
> problem
> down to a change from 'new' to 'basicNew' in OrderedCollection
> class>>new:.  I then tried to find the history of this change and to
> figure out the logic.  To make a long story short the old methods for
> doing this don't work so well in the new 3.9a world and I'm having
> trouble making sense of what I see under Monticello.  From the
> annotation panes it appears that OrderedCollection class>>new: was  
> last
> modified in Collections-md.20.  However the only comment for this mcz
> (we need a new term) is 'add isDictionary' however any way in which I
> look at it shows me many many changes.

> This is just an example, but overall I think its time that we had some
> sort of tutorial session on how to work within the new system.
>

The problem is that a tutorial is not what is needed: New tools are  
needed.
There is no way (that I now) to get from a method to the mcz that
changed that method.

One thing that's clear: The changesets that monticello does are the
diff between the version that is in the image and the one loaded.
So the Collections-md.20 changeset thus does not only show the chanes
of Collections-md.20.mcz

For the basicNew: change: I did make a pass through the system
to make sure that often instanziated system classes don't try to
do an initialize when they don't implement it. (e.g. Point, LookUpKey,
Rectangle and classes like that).

Of course, if someone subclasses these classes with the expectation
  that initialize is called, then this will brake, but is trivially  
fixable.

But subclassing OrderedCollection is not a good idea in any case...

        Marcus