interpreter features

[Simon Peyton-Jones]

| What about my idea? Can't there be a module, say called=20
| *interpreter* which=20
| you can add and remove definitions from on-the-fly?
| Something that would put you tighter in the loop. Without=20
| having to go back=20
| to the editor edit stuff, reload from interpreter, choke on=20
| error, see what=20
| went wrong, etc. Just scratch some code on the CLI, see what=20
| you've defined=20
| so far, edit the current module, and evaluate what you want=20
| to.=20

So the CLI would have to have a way to show you what the=20
current "interpreter" module was, and let you edit it.  How=20
does that differ from an editor?   This smells like a tarpit to
me: once you provide a basic editor, everyone will want more
features.  Better, surely, to use an existing editor?

S