On Jul 19, 2007, at 10:47 PM, skaller wrote: > On Thu, 2007-07-19 at 22:28 -0400, Peter Tanski wrote: >> On 19 July 2007, Erick Tryzelaar wrote: > >> There might be a way to do >> interpreted code with a hook through CINT <http://root.cern.ch/root/ >> Cint.html>. So that could handle code generation and execution; for >> parsing the command line it would take a lot of work, even sponging >> off of readline and the current parser, because each statement would >> have to be wrapped with the current environment. > > Parsing can be done incrementally, and possibly symbol table building. > Everything after that is wholistic. > > Lets face it: Felix doesn't even have separate compilation. > REPL is a refinement of separate compilation -- Felix is > designed as a WPA -- Whole Program Analyser.
One that doesn't require 2GB RAM to run :) > You could do some tricks. For example, accumulate definitions > without rebuilding, since definitions don't "do" anything. > > Also don't forget, Felix uses *setwise* lookup: the scope > of an identifier is the context in which it is defined, > and that includes *previous* statements (like C function scope). With an interpreter you would need to keep a global "interpreter" scope that, in the absence of bytecode, would probably require maintaining a sourcecode state in memory (the parser state) so you could modify it according to the next interpreter statement, output and recompile and relink the whole object again. Personally, I don't think it is worth it: given more time I would rather master felix. You know, until a few emails ago I did not know the 'static' option existed, which is what I wanted since felix 1.1. Cheers, Pete ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Felix-language mailing list Felix-language@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/felix-language