h@l@h@!h@C+h!k1k0-h@$k h@k1k0-+$h@C+h!ih@!h@C+h!kefh@!h@C+h!l! h@l@h@!h@C+h!k1k0-h@$k h@k1k0-+$h@C+h!ih@!h@C+h!kefh@!h@C+h!l! h@l@ h@!h@C+h! k1k0-h@$ k\h@k1k0-+$ h@C+h! i h@!h@C+h! kkf h@!h@C+h! kLf h@!h@C+h! k:k0- h@!h@C+h! k=f h@!h@C+h! kJf h@!h@C+h! k0k5-C* h@!h@C+h! kef h@!h@C+h! l! \ **Now we can use single-line comments!** \ planckforth - \ Copyright (C) 2020 nineties \ This project aims to create a full-fledged Forth interpreter \ by bootstrapping from hand-written tiny ELF binary. \ In the 1st stage, only single character words are registered \ in the dictionary. \ List of builtin words: \ 'Q' ( -- ) Exit the process \ 'C' ( -- n ) The size of Cells \ 'h' ( -- addr ) The address of 'here' cell \ 'l' ( -- addr ) The address of 'latest' cell \ 'k' ( -- c ) Read character \ 't' ( c -- ) Print character \ 'j' ( -- ) Unconditional branch \ 'J' ( a -- ) Jump if a == 0 \ 'f' ( c -- xt ) Get execution token of c \ 'x' ( xt -- ... ) Run the execution token \ '@' ( addr -- a ) Load value from addr \ '!' ( a addr -- ) Store value to addr \ '?' ( addr -- c ) Load byte from addr \ '$' ( c addr -- ) Store byte to addr \ 'd' ( -- addr ) Get data stack pointer \ 'D' ( addr -- ) Set data stack pointer \ 'r' ( -- addr ) Get return stack pointer \ 'R' ( addr -- ) Set return stack pointer \ 'i' ( -- addr ) Get the interpreter function \ 'e' ( -- ) Exit current function \ 'L' ( -- a ) Load immediate \ 'S' ( -- addr len) Load string literal \ '+' ( a b -- c ) c = (a + b) \ '-' ( a b -- c ) c = (a - b) \ '*' ( a b -- c ) c = (a * b) \ '/' ( a b -- c ) c = (a / b) \ '%' ( a b -- c ) c = (a % b) \ '&' ( a b -- c ) c = (a & b) \ '|' ( a b -- c ) c = (a | b) \ '^' ( a b -- c ) c = (a ^ b) \ '<' ( a b -- c ) c = (a < b) \ '=' ( a b -- c ) c = (a == b \ The 1st stage interpreter repeats execution of k, f and x. \ There is no concept such as IMMEDIATE mode yet. \ There following line is an example program of planckforth \ which prints "Hello World!\n" \ -- \ kHtketkltkltkotk tkWtkotkrtkltkdtk!tk:k0-tQ \ -- \ This code repeats that 'k' reads a character and 't' prints it. \ Note that ':' (58) minus '0' (48) is '\n' (10). \ The structure of the dictionary is sames as many Forth implementation. \ +------+----------+---------+------------+---------------+ \ | link | len+flag | name... | padding... | code field ...| \ +------+----------+---------+------------+---------------+ \ - link pointer to the previous entry (CELL byte) \ - length of the name (5 bits) \ - immediate flag (1 bit) \ - smuege flag (1 bit) \ - unused bit (1 bit) \ - characters of the name (N bits) \ - padding to align CELL boundary if necessary. \ - codewords and datawords (CELL-bye aligned) \ The code group at the beginning of this file \ defines ' ' and '\n' as no-op operation and \ '\' to read following characters until '\n'. \ Since I couldn't write a comment at the beginning, \ I repost the definition of '\' for explanation. \ -- \ h@ ( save addr of new entry ) \ l@ h@!h@C+h! ( set link pointer. *here++ = latest ) \ k1k0-h@$ k\h@k1k0-+$ h@C+h! ( write the name '\' and its length ) \ i h@!h@C+h! ( docol ) \ kkf h@!h@C+h! ( key ) \ kLf h@!h@C+h! ( lit ) \ k:k0- h@!h@C+h! ( '\n' ) \ k=f h@!h@C+h! ( = ) \ kJf h@!h@C+h! ( branch ) \ k0k5-C* h@!h@C+h! ( -5*CELL ) \ kef h@!h@C+h! ( exit ) \ l! ( set latest to this new entry. ) \ -- \ That's all for the brief explanation. Let's restart bootstrap! \ The COMMA operator \ ',' ( a -- ) Store a to 'here' and increment 'here' CELL bytes. h@l@ h@!h@C+h! k1k0-h@$ k,h@k1k0-+$ h@C+h! i h@!h@C+h! \ docol \ store 'a' to here khf h@!h@C+h! k@f h@!h@C+h! k!f h@!h@C+h! \ here <- here + CELL khf h@!h@C+h! k@f h@!h@C+h! kCf h@!h@C+h! k+f h@!h@C+h! khf h@!h@C+h! k!f h@!h@C+h! \ exit kef h@!h@C+h! l! \ TICK-like operator \ '\'' ( "c" -- xt ) Get execution token of following character \ NB: This definition is different from the usual definition of tick \ because it does not skip leading spaces and can read only a single \ character. It will be redefined in later stage. h@l@, k1k0-h@$ k'h@k1k0-+$ h@C+h! i, kkf, kff, kef, l! \ Utility for defining a word \ 'c' ( "c" -- w ) \ Read character, create new word then push its address. \ 'latest' will not be updated. h@l@, k1k0-h@$ kch@k1k0-+$ h@C+h! i, 'h, '@, 'l, '@, ',, 'L, k1k0-, 'h, '@, '$, 'k, 'h, '@, 'L, k1k0-, '+, '$, 'h, '@, 'C, '+, 'h, '!, 'e, l! \ '_' ( a -- ) DROP c_ i, 'd, 'C, '+, 'D, 'e, l! \ '#' ( a -- a a ) DUP c# i, 'd, '@, 'e, l! \ Implementations of TOR and FROMR are a bit tricky. \ Since return-address will be placed at the top of return stack, \ the code in the body of these function have to manipulate \ 2nd element of the stack. \ '{' ( a -- R:a ) TOR \ Move value from data stack to return stack. c{ i, 'r, 'r, '@, \ ( a rsp ret ) 'r, 'C, '-, '#, \ ( a rsp ret rsp-1 rsp-1 ) 'R, \ ( a rsp+1 ret rsp ) extend return stack '!, \ ( a rsp+1 ) store return address to the top '!, \ store a to the 2nd 'e, l! \ '}' ( R:a -- a ) FROMR \ Move value from return stack to data stack. c} i, 'r, 'C, '+, '@, \ ( a ) load 2nd value 'r, '@, \ ( a ret ) load return addr 'r, 'C, '+, '#, \ ( a ret rsp+1 rsp+1 ) 'R, \ ( a ret rsp ) reduce return stack '!, \ ( a , R:ret ) store return addr to top of return stack 'e, l! \ 'o' ( a b -- a b a ) OVER co i, 'd, 'C, '+, '@, 'e, l! \ '~' ( a b -- b a ) SWAP c~ i, 'o, \ ( a b a ) '{, \ ( a b , R:a ) 'd, 'C, '+, \ ( a b sp+1 , R:a ) '!, \ ( b , R:a ) '}, \ ( b a ) 'e, l! \ 'B' ( c -- ) C-COMMA \ Store byte 'c' to here and increment it cB i, 'h, '@, '$, 'h, 'L, k1k0-, '+, 'h, '!, 'e, l! \ 'a' ( c-addr -- a-addr ) ALIGNED \ Round up 'a' to a multiple of CELL ca i, 'L, Ck1k0--, '+, \ ( a+CELL-1 ) 'L, k0k0-C-, \ ( a+CELL-1 ~(CELL-1) ) '&, 'e, l! \ 'A' ( -- ) ALIGN \ Round up 'here' to a multiple of CELL cA i, 'h, '@, 'a, 'h, '!, 'e, l! Q