ByteCodePrimitives

 \ 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.