Em 28/05/2014 08:58, Byron Jeff escreveu: > Actually this sort of changes my mind on it (which is why I prefaced the > entire statement with "may or may not" to begin with...). Since tokens we= re > in a different space than values, it's trivial to determine that the firs= t > token after an end is a variable if the top bit is not set.=20 > > That would make the LET redundant, and clearly occupying valuable space. > > BTW I too wish that I had come across FORTH early on instead of 6 or so > years ago. I'm in the process of putting together a completely self hoste= d > FORTH development system for the PIC24F series of chips. It would have > saved me the trouble of trying to write my own tokenized C-like > cross-compiler oh so many years ago. I too am in a pursuit for some sort of interpreted/scripting language for embedded systems for a long time. My motivation it to allow the users of my boards to customize their operation or write their own routines. The problem is that I don't want them to be able to run native code because it would allow someone to steal my own code. That protection could be done with dsPIC's CodeGuard, but it is not available for other platforms, and I don't know how secure it is indeed. Some years ago, I tried XPL0 (). It is a language created in 1976 by Peter J. R. Boyle and maintained by Loren Blaney. There is a compiler that generates 80x86 code and another that generates a psudo-code called I2L. The original I2L interpreter was written in 80x86 assembly and I wrote a very portable new one in C to use in my boards. The performance was acceptable, although not great. It worked, but I wasn't satisfied with the fact that it is one more language to be learned and it is very obscure. That would prevent its adoption by my customers. I'm working now in a Thumb emulator written in C, that can be compiled for a lot of platforms (even 8-bit, although the performance would be awful). This emulator creates a virtual ARM processor (Thumb1 instructions only) that won't allow any access outside its private memory pool. System calls, common library functions and other predefined operations are implemented natively and called with the Thumb "svc" instruction. I managed to convince the ARM-GCC to generate code for this virtual processor and it seems very promising. The user's applications can be written in any language with a compiler that generates Thumb code. Right now I don't have time to polish it, but I'm excited about the possibilities. Isaac --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .