>if you make
>things clear on your marketing efforts that the focus is at embedding and having
>a lightweight, 100% ANSI-C core, there's no reason not to start developing
>fancy add-ons around it.
>And if you are really concerned abount mixing things up, then develop the add-on
>at a separate project-website - get a sourceforge account for it, etc.
This is exactly what I mean. The headline on the Lua website should be "here's a language designed for embeddability (and by the way, some of the things you can embed it in are a command line tool, an IDE, etc)" in contrast to "here's another language with an IDE, a GUI toolkit, a standard library, etc (oh and by the way, you can embed it if you like)" which is what you get with other languages. Put the emphasis on embedding and reinforce and promote the language in this area and you have a distinctive identity.
Deep down, programmers know that you can use almost any language for almost anything, but still tend to associate different languages with different types of task (Lisp is seen as an AI language, Perl is seen as a text processing language, etc). All this means is that the language has a slight bias toward the task it is associated with, but the difference it makes from a marketing point of view is enormous. The difficulty Lua faces is that its target task of extreme embeddability is relatively unexplored territory. I think that injecting a few novel ideas about embeddability into Lua will help people to see that it is not a trivial task and deserves special treatment.