Just a reflexion that I��d like to share here.

When Hilaire talks about complexity. I have to agree too, even if I like the progress under the hood, and explored spaces like GToolkit.

But how to change such legacy beast, the pharo image ? Potential is huge but resistance is also tough for so few people�� Just imagining changing method protocols seems so complex at the end !

Instead, I��d start with minimal images, bootstrapped ones. A system I would love to understand every bits. I��m pretty sure this is what lots of us have in mind and the progress in bootstrap has been teasing us ;-)

For instance, I��d want minimal images that would be able to deal with collections of informations (text/data processing) with (web)socket connectivity or whatever cheapest way to communicate. It would be some kind of information store, a minimal image that can store information objects and that I can interact with a client in a full pharo image (or in a web browser).

Wouldn��t it be possible to start from scratch, literally from nothing (nil) and define little by little what necessary and along the way define minimal protocols, minimal Object class and then iteratively a clean collection hierarchy until �� (for me) �� the information store.

That would probably need some filesystem support but it��s not an obligation. I��d far prefer web socket stream support and interaction with object only within the image. Another image could be responsible of exporting information objects to files�� It also need some basic mime types support and probably parsing (xml/html, json, geojson, ical, ��). I don��t know well. But it has to be iteratively done anyway.

I imagine there are not one but several images (and therefore source code) that would lead to functionally equivalent images. There should be some ways to abstract functionalities we expect of a system (if people have pointers on that) ?

We could even imagine like a Contest to come up with the nicest pharo images (contest driven by tests ^^).

Depending on the functionality coverage, that��d be just crazy to see similarities between code/images��

</dreamMode>

Is it feasible, does some of the remarks makes sense to some of you ?

At least, Caro, as we talked a bit, your writings on such a quest will be terrific, and I��m sure well followed ;-)