( "name" -- )
ALSO and add ASSEMBLER wordlist if available. Add PROC ENTER
assembler snippet as needed for the architecture into the PFA. The
CFA is setup (a) with the PFA adress in traditional ITC or (b)
with an infoblock as for sbr-coded colon words.
Remember that not all architectures are support and that the
ASSEMBLER wordset is not compiled into pfe by default. Use always
END-CODE for each
CODE start. The new
word name is not smudged.
primitive code = [p4_asm_create_code]
( -- )
Does end the latest word (being usually some DOES> part) and enters
machine-level (in EXEC-mode).
The TOOLS-EXT will not provide an
END-CODE or any other word in the
ASSEMBLER wordlist which is required to start any useful assembler
programming. After requiring ASSEMBLER-EXT you will see a second ";CODE"
EXTENSIONS wordlist that will also provide an optimized execution
than the result of this standard-forth implemenation.
The Standard-Forth implementation will actually compile a derivate of
BRANCH into the dictionary followed by
;. The compiled word
will not jump to the target adress (following the execution token)
but it will call the target adress via the host C stack. The target
machine level word (C domain) will just return here for being
returned (Forth domain). Hence
END-CODE may be a simple RET, comma!
primitive code = [p4_asm_semicolon_code]
( "name" -- )
PREVIOUS and add PROC LEAVE assembler snippet as needed
for the architecture - usually includes bits to "return from
subroutine". Remember that not all architectures are support and
PFE usually does only do variants of call-threading with a separate
loop for the inner interpreter that does "call into subroutine".
Some forth implementations do "jump into routine" and the PROC
LEAVE part would do "jump to next routine" also known as
next-threading. The sbr-call-threading is usually similar to the
native subroutine-coding of the host operating system. See
primitive code = [p4_asm_end_code]