Monticello status

Avi Bryant avi at beta4.com
Mon Apr 7 08:25:29 UTC 2003


On Mon, 7 Apr 2003, Stephane Ducasse wrote:

> I think that this is really similar, the key question is what is the
> granularity we want. For example should we have instance variable
> object.

Yes, see my reply to your other message.

> We should also be able to annotate any entity with information that has
> not been upfront thought of. So having a dictionary where not
> predefined properties are stored
> is good.

Yes, though this has the same problem as separate instance variables -
it's not that useful until the tools know about it too.  Nobody's going to
add metadata by hand to a fileOut, and even if they did it wouldn't be
"round tripped" - loading the code in and filing it out again would lose
the annotations.

> Do you have some tests about what you have now?

Yes.  They mostly have to do with diff/merge semantics, but there are also
some tests for loading declarations from disk<->model and from
model<->image.

Avi




More information about the Squeak-dev mailing list