> Well, no. Broken code is bad. ;-)
>
> But still, I stand by that comment in the context it was written. I 
> was talking about dependencies in Monticello, not SqueakMap. The two 
> are very, very different things.
>
> I've made the point before, and I'll make it again: Monticello is 
> *not* a packaging system. It's a versioning system. Yes, it does 
> version packages, and it does have some convenience features for 
> publishing them, but building an image by locating, downloading and 
> installing packages that don't conflict is SqueakMap's job.
>
> So why does Monticello track dependencies? It's so that "applications" 
> can be factored into several packages, while still being developed and 
> versioned together. So, for example, when I load SqueakSource into my 
> development environment, I know that I'm also getting the exactly 
> version of Mewa that it was developed against. And since both are 
> right there in my repository, that's not a problem.
>
> SqueakMap is a different beast. When loading a package from SM, you 
> don't need an exact match for the dependency. You just need one that 
> works. I think Göran's scheme for figuring that out is quite elegant, 
> and I can't wait until we have it up and running. Until then, we'll 
> just have to make do without dependencies in SqueakMap. Using 
> Monticello dependencies instead will only lead to problems like the 
> one Hernan ran into.
>
> Stéphane, this rant wasn't directed entirely at you. For some reason 
> this issue seems to come up quite often, and produces confusion every 
> time.
No problem
I just want to see that working because software is complex. Then if 
Smacc depends on RB. When I load SmaCC1.23 and it needs
to work RB 3.5. I should get that.

Stef