\ Here is a sample set of primitives. These were found by taking the \ Standard Forth core and some core extension words, and throwing out \ the ones that cause trouble. \ T: is the name of a command that assigns bytecodes to these words \ on the tokenizer. The tokenizer also defines additional tokens \ that require special treatment. \ T: is also the name of a different command that assigns execution \ tokens to byte tokens on the token compiler. The token compiler \ has an additional T: ;T sequence to handle the special words that \ didn't fit into this list. \ T: is also the name of a third command that assigns names to tokens \ on the detokenizer. There is a third T: ;T sequence to give new names \ to the special words. \ So to make a new byte-code system, you might start by noticing which \ words you need, and create a list like this. Then you notice \ which words require special attention. Those will be particularly \ parsing words, compiling words, and defining words. They will each need \ special treatment, probably in both the tokenizer and the token-compiler. \ The words that don't need special treatment can go into this list. T: ! # #> #S * */ */MOD + +! +LOOP , - . / /MOD 0< 0= 1+ 1- 2! 2* 2/ 2@ 2DROP 2DUP 2OVER < <# = > >BODY >IN >NUMBER >R ?DUP @ ABORT ABS ACCEPT ALIGN ALIGNED ALLOT AND BASE BEGIN BL C! C, C@ CELL+ CELLS CHAR+ CHARS COUNT CR DO DECIMAL DEPTH DROP DUP ELSE EMIT EXECUTE EXIT FILL FM/MOD HERE HOLD I IF IMMEDIATE INVERT J KEY LEAVE LITERAL LOOP LSHIFT M* MAX MIN MOD MOVE NEGATE OR OVER QUIT R> R@ RECURSE REPEAT ROT RSHIFT S>D SIGN SM/REM SOURCE SPACE SPACES STATE SWAP THEN TYPE U. U< UM* UM/MOD UNLOOP UNTIL WHILE WORD XOR .S ;T \ WORD is the only one above that parses the input stream. \ WORD should only be compiled into commands that will be executed \ on the target, and that won't be expected to act on byte-code. \ Since the byte-code is not a character input stream, WORD won't work on it. \ Words that parse and words that compile will often need special treatment \ in the tokenizer or the token-compiler or both.
