Hello everybody. I was playing with Haskell and particularly Hugs lately and came to the conclusion that Hugs-style interpreter can be a real blessing as embedded application language (akin to ecmascript interpreters or guile). Two nicest features in this direction are: 1. The ability to evaluate arbitrarily complex expressions interactively without the ability to redefine names. 2. Simple expressions in Haskell are more intuitive than most languages offer. I understand that I've just invented the wheel. However, after looking at hugs source code and hugs server api I think few very desirable features shall be implemented: 1. api to control importing of modules 2. api to control resolution of foreign symbols 3. Additionally it would be nice to have the ability to evaluate literal Haskell expressions from within Haskell program, allowing Haskell implementations of preprocessors for user defined language features. Also, in its current state Hugs ability to run programs is quite lacking in certain areas (I mean resource management - memory and stack). This calls for next point: 4. No internal limitations in the interpreter. Let it fail only when system resources are exhausted. And two more points: 5. Let interpreter/compiler to exist separately from possible frontends in form of library. 6. It will be nice to have support for precompilation of modules. I have a great desire to make an implementation of Haskell interpreter with all these features. I want to do it in C++, as this will allow much cleaner implementation than C (no more macros!) without large performance loss (not that it matters greatly). I, however, need a help with top-level design of the interpreter. Does anybody has any suggestions on data structures and code organization in general? May be some work was done already in the direction that can be useful? I'll appreciate any feedback to this message. ____________________________________________________ Start your day with Yahoo! - make it your home page http://www.yahoo.com/r/hs