[PicForth] Problems/solutions with ftables

[David P Harris]

David-
For a proper solution to ables going over a page boundary, have a look 
at the code provided by the PICLIST contributors at 
http://www.piclist.com/techref/microchip/memory.htm
Then you con have tables anywhere and larger than 256 bytes.  Also, it 
would be nice to define a Build routine that would let you make your 
strings in the source code in readable form. 
David

David McNab wrote:

> Hi,
>
> Just a few quick observations:
>
> 1. I've noticed that program-memory tables, created with 
> ftable...end-table, only work if all the data in the table resides in 
> the first 256 bytes of program memory. The ADDWF PCL,F trick used for 
> indexing the table seems to set PCH to $00, regardless of the current IP.
>
> 2. If file1.fs includes file2.fs, and file2.fs contains an 
> ftable..end-table construct, subsequent includes from file1.fs fail.
>
> 3. libstrings.fs doesn't work. Specifically, str-char is always 
> returning $01. I've used exactly the code from the doco for 
> 'c".."'/'str-char'. No good.
>
> Anyway, here's what I've done to work around these issues:
>
> I'm working on a program that does lots of user interaction via 
> buttons and LCD, and requires several strings to be stored. So I've 
> chosen to put them in program memory.
>
> What I've had to do is write a utility that automatically converts a 
> file containing something like:
>
>     fred = "hello, world"
>     mary = "smaller"
>     jim = "blah"
>
> into:
>
>   ftable mystrings
>   $68 t, $65 t, $6c t, $6c t, $6f t, $2c t, $20 t, $77 t,
>   $6f t, $72 t, $6c t, $64 t, 0 t, $62 t, $6c t, $61 t,
>   $68 t, 0 t, $73 t, $6d t, $61 t, $6c t, $6c t, $65 t,
>   $72 t, 0 t,
>   end-table
>   $0 constant fred
>   $d constant jim
>   $12 constant mary
>
> Here, 'fred', 'jim' and 'mary' are offsets into this table, pointing 
> to the start of null-terminated strings.
>
> This allows any chosen offset to be passed to a loop, hardcoded to use 
> 'mystrings', which indexes this table and invokes a callback for each 
> char in the string.
>
> Finally, I've arrived at a way to enter strings in human-readable 
> format, have them automatically translated to table decs, such that I 
> can do stuff like:
>
>   lcd-line1 fred lcd-put-string
>   lcd-line2 mary lcd-put-string
>
> And as long as I ensure that the whole table is residing in first 256 
> bytes, all is ok. What's really great about ftable is that it takes 
> advantage of the RETLW instruction, so that there is zero overhead for 
> each string (apart from its null-terminator).
>
> (Makes me wonder if i should hack up an encoder to store 2 7-bit ASCII 
> chars in each machine word).
>
> All this said, I find that overall, PicForth is a god-send for my PIC 
> project. Between it and gpsim, I'm in PIC Hacking Heaven, especially 
> after weaning myself off the insulting PICAXE BASIC environment. So 
> thank you, Samuel, for a master-work! :))
> I really hope you find the inspiration to keep going on it.
>