Göran wrote:

> So... I want a smart Smalltalk diff/patch but for *objects* and not
> Strings. Anyone up to the task? Magma should in essence already have
> this capability somewhere in there I think. :)

I replied:

> Well, I've been working on something a bit like this for Monticello.
> The thing is, a general diff/patch algorithm for arbitrary graphs of
> objects is really, really hard..

Chris concurred:

> An interesting, very tough problem..  Too tough for me.

And finally, Stephan Rudlof wrote:

> Some way to go is to look for XML diff algorithms. They compare XML 
> files.
>
> A Smalltalk class hierarchy can be exported as XML:
> - classes
>   - name
>   - superclass name
>   - inst var names
>   - method names
>   - method bytecode hashes (!)
>   - optional: method source (not so important than the previous point)
> ...
> After that these XML exports could be diffed.
>
> Of course the diff problem is hard and the patch problem is harder than
> that: there are more than one ways to convert one XML graph into 
> another...
> But the benefit is to get an idea of
> - how far two different Smalltalk systems are away from each other,
> - *where* they differ,
> - and - last but not least - how much the porting effort of some
> software could be ... (!)

Hi Stephane,

This is more or less the approach I have taken for Monticello. XML has 
spurred research into tree-diffing algorithms, since straight text 
diffs only sort-of-work for XML. The various algorithms I've run across 
all make different trade-offs; the one I've selected aims for simple, 
and can be adapted to the "ugly" trees that Smalltalk source-code 
forms, but is quite complex and (possibly) slow on really large trees.

On the other hand, this is a simplified version of the general problem 
that Göran and Chris were talking about - the ability to diff and patch 
any graph of objects, not just source code (which, luckily, can be 
restricted to a tree without many problems.)

Colin