Morph cleaning proposal

diegogomezdeck at diegogomezdeck at
Thu Feb 13 14:14:20 UTC 2003

Thanks for answer!

We're on the road:


Diego Gomez Deck

> Hi Diego . I think it's a good direction.
> Since I want to see your work used as much as you do, and since nobody
> can give you any sort of warranty, let's do it another way - I'll give
> you some tips that minimize your risk. It may seem less exciting this
> way, but I'm talking from experience here - there's nothing more
> frustrating in this arena than running ahead implementing what seems
> enough to the implementor and finding that the people that need to
> adopt it don't see it the same way.
> This is relevant to anyone (and not just in the Squeak world).
> * Don't make your changes dependent on any technology that isn't
> already in the image. If you do, don't hope for them to become anything
> more than an SM package. It's better to scale down the project or delay
> it, because what seems like shoo-in for adoption today, might end up
> taking a couple of months or years yet. Just ask Craig, Anthony,
> Henrik... So for now, I strongly suggest you not depend on Traits.
> * Your idea of taking small pieces is great. Under the new rules of the
> game, it's almost a requirement. Before your stuff gets into the image,
> it will have to be reviewed by quite a few people, and before they put
> that time into it, they'll want to understand what it's good for. So
> you want to structure your work as many small bits *that are each
> obviously a good thing by itself*. If you're doing refactorings, the
> same idea applies. Don't change everything around - make local changes,
> and let the add up.
> * As you can see from the recent discussion, radical changes are
> controvertial. We're probably not ready to do something as big as
> convert Morphic to Traits, or separate it completely from eToys. We
> don't yet know enough. Start small. Start with obviously good things.
> Want an example? Make Genie separatable from Morphic. If you want more
> likely tasks, I'll generate more.
> * Build the work process over time. Send in something really small at
> first, we'll have a look. We'll have comments. As we get our
> expectations synchronized, you'll be able to assess what we'll accept
> and what we won't. And then you can decide to do bigger pieces (take a
> bigger chance).
> Does this make sense to you?
> Daniel Vainsencher (Guide in charge of refactoring the image, along
> with Ned)

More information about the Squeak-dev mailing list