Add UNRAVEL from HMSL

[philburk]

Implementation at:

https://github.com/philburk/hmsl/blob/cf49471aee4543ae4dd045e5bae4693a7629b08e/hmsl/fth/errormsg.fth#L22

[robzed]

I have currently have -fsanitize=address turned on in my pForth build and I immediately get a "ERROR: AddressSanitizer: heap-buffer-overflow" - I assume because it's walking into memory below the return stack with the 20 0 do. I assume we'd need to have a R0 or RP0 command and use the commented out code only to print out the used return stack area?

[robzed]

When I test (with unravel inside LOAD_GRAPHICS) I get this:
run
Calling sequence: LOAD_GRAPHICS RUN 0
=================================================================
==29158==ERROR: AddressSanitizer: heap-buffer-overflow

[robzed]

From the command line:

rp@    ok
Stack<10> 107820859018496 

From the command line:

: tt1 rp@ . 3 >r 4 >r  rp@ . R> r> drop drop ;    ok
Stack<10> 
tt1 107820859018488 107820859018472    ok

Actually I prototyped it in a nested include with:
rp@ value rp0

Gives:

Stack<10> 
rp0 . 107820859018416    ok
Stack<10> 

Inside first include:

." Default stack = " rp0 . cr

Default stack = 107820859018456

[robzed]

Inside top level include
rp@ 40 + value rp0

(the 40 + is to adjust for the fact this is defined in an include. In my opinion, this should be a primitive).

Anywhere else afterwards:

: UNRAVEL  ( -- , show names on stack )
    >newline ." Calling sequence:" cr
    rp0 rp@ - cell / 
    ." (Levels = " dup . ." )" cr    ( added for test purposes )
\    2+
   50 min 0
\    20 0
    ?DO  4 spaces
        rp@ i 2+ cell* + @
        dup code> >name ?dup
        IF id. drop
        ELSE .hex
        THEN cr?
    LOOP cr
;

Gives:

run 
Calling sequence:
(Levels = 3 )
    LOAD_GRAPHICS    RUN    0 

No crash.

Some questions:

• Should the number of levels be 2?
• Why is it printing the 0?
• What was the 2+ I commented out for?
• What is the 2+ in the loop for?

[philburk]

Thanks for experimenting with this.
I agree that RP0 or R0 should be in the kernel.
In JForth we defined R0
http://www.jforth.org/Glos_RT.html

I cannot find R0 or RP0 in the ANS standard. Do you know of any standard references?
FIG Forth defined R0.

Should the number of levels be 2?

Probably. I need to ponder this.

Why is it printing the 0?

Not sure. It may be from the DO LOOP

What was the 2+ I commented out for?

I think that was from some old JForth code.

What is the 2+ in the loop for?

I think it is to get past the DO LOOP internal values and the call to UNRAVEL.

[philburk]

@robzed - please take a look at #197

[robzed]

Happy to merge. Did you want me to do more on this?