Re: [Flashforth-devel] FlashForth 2-Dimensional arrays?

[<dab...@ya...>]

 Very Nice Thought! 
(As the War College replied to Grace Murray Hopper about her answer on the problem of refueling a battle group at sea, "An interesting solution.")
I'm going to have to give that one a try once this project is wrapped up. Much cleaner than anything I came up with. Glad I'm not the only one who couldn't see a way back to the DOES>... (grin)
Thanks Again!
Keep up the good work. Stephen's warnings aside, it pays to know the underlying system.

     On Monday, June 3, 2019, 9:35:57 AM EDT, Mikael Nordman <mik...@fl...> wrote:  
 
 Sorry Craig,

I cannot think of any good way to do that in FF.

Once you are executing in the DOES> part, which part of the 2ARRAY word,
I cannot think of any way to get back to defined array word to fetch a 
size from flash.
That address is gone from the return stack.
Storing the size in flash can be done in the CREATE part with I, or 
FLASH , RAM
But when the DOES> part is executing, there is no link to the array 
word.

The only way I can think of is to initialize the size of each array in 
ram after every warm start.

Except, there is a way :-) Not tested or complete, but you get the idea.

: 2array ( col, row -- addr )
  \ create a word which points to a data area in flash
  flash create
  \ append the row size in flash
  dup ,
  \ append the start of the ram buffer in flash
  ram here flash ,
  \ allot the the ram buffer
  cells * ram here allot
  does>
    \ Fetch the row size from flash
    @+
    \ Fetch the ram buffer address flash
    @
    \ etcetera
  ;

BR Mike

On 2019-06-02 16:55, craig bair via Flashforth-devel wrote:
> I've run into an interesting problem with FF-5 on Pic18F46K22 chips.
> 2-Dimensional arrays are really handy things.  I'm setting up a
> rolling-average filter for the output of a multi-channel A/D chip to
> take a reading from a different channel each millisecond and average
> the reading I need from the correct ring-buffer whenever needed.  The
> control routine works perfectly with a 2D-array in Forth to the limits
> that I can test it... until I shut the system down or do a reset.
> Suddenly the 2array word stops returning addresses other than for row
> Zero.
> 
> Here's the word I use, derived from Stephen Pelc's (cleaned up so it
> doesn't crash when accessed):
> \ WORD ARRAY CREATION -- 2-Dimensional ( col, row -- addr )
> \ x0y0 x1y0 x2y0 ... xny0 x0y1 x1y1 ... xny1 x0y2 ... xnym
> : 2array    \ x1 y2 -- ; [child] n1 n2 -- addr ; 2-D array
>        \ make name word, save width, calculate size, allocate
>    create over , cells * here over erase allot
>        \ run time gives address of data
>    does> dup @ rot * rot + 1+ cells +
> ;
> 
> My best initial guess is that the COMMA is storing the row length into
> a ram location so a reset clears it to zero from that point on (in a
> Harvard Architecture processor).
> 
> I can work around it by doing a specific address calculation routine
> with a 1D-array, but fixing "2array" for embedded controller would be
> better.  Any suggestions or insights?
> 
> craig
> 
> 
> 
> 
> _______________________________________________
> Flashforth-devel mailing list
> Fla...@li...
> https://lists.sourceforge.net/lists/listinfo/flashforth-devel

-- 
--
Mikael


_______________________________________________
Flashforth-devel mailing list
Fla...@li...
https://lists.sourceforge.net/lists/listinfo/flashforth-devel