Fix [ELSE] reference implementation to handle comment lines with conditional compilation words.

[SirWumpus]

The Forth Draft 19.1 reference implementation for [ELSE] (section E.15.6.2.2531) should be improved to allow for single-line commented [IF], [ELSE], and [THEN] words. I stumbled on this cause I had commented out some conditional compilation words while experimenting / tweaking some code and started to notice some odd behaviour.

Consider the following revised test cases:

\ Multiline with single line comment of [IF] [ELSE] [THEN] words, ie.
\ developer is futzing around with code and temorarily comments words.
T{ TRUE [IF]
    \ TRUE [IF]
        1
    \ [ELSE]
        2
    [ELSE]
        3
    \ [THEN]
        4
    [THEN] -> 1 2 }T

T{ FALSE [IF]
        1
    \ [ELSE]
        2
    [ELSE]
    \ TRUE [IF}
        3
    \ [THEN]
        4
    [THEN] -> 3 4 }T

The following revised [ELSE] solves the above:

: [ELSE] ( -- )
	1 BEGIN 				\ level
	  BEGIN PARSE-NAME DUP WHILE		\ level adr len
	    2DUP S" \" COMPARE 0= IF
	      \ Ignore remainder of comment line.
	      2DROP POSTPONE \
	    ELSE 2DUP S" [IF]" COMPARE 0= IF		\ level adr len
	      2DROP 1+				\ level'
	    ELSE				\ level adr len
	      2DUP S" [ELSE]" COMPARE 0= IF	\ level adr len
	        2DROP 1-			\ level'
	        \ Not yet zero, then restore previous level while nested.
	        DUP IF 1+ THEN			\ level'
	      ELSE 				\ level adr len
	        S" [THEN]" COMPARE 0= IF	\ level
	          1-				\ level'
	        THEN
	      THEN
	    THEN THEN
	    ?DUP 0= IF EXIT THEN		\ level'
	  REPEAT 2DROP				\ level
	REFILL 0= UNTIL				\ level
	DROP
; IMMEDIATE

Note the above does not handle parenthesis comments (which I tend not to use if possible).

[ruv]

This problem cannot be solved in Forth.

A user can redefine line comment (or introduce an own variant):

: \G   \ special handling ... 
  postpone \
; immediate

Or [else] can be used in a string literal like s" .... [else] ... " inside a containing [if] [then].

It's confused if these words can be commented by \, but things become broken if they are appeared in other comments or string literals.

So they should be considered as a top-level markup, which should not appear in comments or string literals.

This problem was also discussed on forth-standard.org.

[SirWumpus]

Ah. Well the only reason I had to comment these words out was for temporary experimentation and testing (which is how I stumbled on the issue) and then either uncomment or remove them once a course of action was decided on. I'm just used to how in C I can comment out #ifdef ... #endif lines and insert temporary replacements while toying with the code. Its not something I'd leave in a final working solution.

As with all things Forth, if you redefine standard / common words, you've probably gone off the reservation on some tangent and to expect there be dragons.

Reading that discussion goes to show I'm not the first, but could have been the last if this had been "clarified" in the rationale so as to avoid the mine field.