Indeed why not cleaning it on the way too...
What would be good is to avoid to have to parse method body to get the 
method and args of the method and to be able to add any meta 
information (time, author) and not using a rigid format.
This is one weakness of the chunk format.

Still with the following approach if you have the declarative object 
you can load a code without installing it. and if you want to have that 
with the chunk format for class you have to introduce
PseudoClass too. So the question is really what should be the 
precondition we want in terms of code representation.

On Tuesday, July 29, 2003, at 04:18 PM, Lex Spoon wrote:

> Stephane Ducasse <ducasse at iam.unibe.ch> wrote:
>
>> For me I would have no problem to have:
>>
>> MethodDefinition
>> 	inClass: #MyFooClass;
>> 	inPackage: #TheFooPackage;
>> 	methodName: #foo:bar:;
>> 	arguments: #(one two);
>> 	fullName: 'foo: one bar: two';
>> 	mainComment: "jkljljlkjl";
>> 	body: 'jlkjlkjkj lkjkl j lkj ljkl ';
>> 	timestamp: '10/02/2003';
>> 	version: '1.02';
>> 	author: '....'
>>
>
> A wonderful thing about this approach ties in with the issue Andreas 
> was
> talking about: what happens for people who don't have the package
> management system loaded?  With this kind of approach, the answer can 
> be
> that you *execute* the code.
>
> In fact, you can do so with the existing chunk format.  You just need 
> to
> add some new kinds of chunks.  This seems like the best thing to do all
> around.  It legacy systems file in as usual, and perfectly well lets
> people parse it when that is what they prefer.  It also already exists,
> for the most part.  We just need some chunks for things like "declare
> global variable".  :)
>
> So why not just stick with the current fileout format, and add new 
> kinds
> of chunks as necessary?
>
> Lex
>