Re: Extending Applesoft with Ampersand routines.

[Stavros Karatsoridis <skaratso@gmail.com>]

On Mar 12, 4:48 pm, BLuRry <brendan.rob...@gmail.com> wrote:
> On Tuesday, March 12, 2013 3:32:35 PM UTC-5, Michael J. Mahon wrote:
> > BLuRry <brendan.rob...@gmail.com> wrote:
>
> > > I'm starting this as a new thread since Bill Buckels asked a very good
>
> > > question as a response to my smart-ass response to a cross-post.
>
> > > Here is what I know about ampersand routines from a high-level:
>
> > > You register your ampersand routine handler as a memory pointer, and your
>
> > > routine is called by the applesoft interperter (only at runtime) when it
>
> > > encounters the & sign.
>
> > > Applesoft pointers indicate the current memory address where the parser
>
> > > was looking (I can't remember if it is the location where & occurs or the
>
> > > memory location just after it).
>
> > > It is up to your routine to parse as much or as little as it wants, and I
>
> > > believe your parsing routine has to update the applesoft pointer(s) as
>
> > > well.  This allows you to do some very wonderful and complex things, or
>
> > > also very simple ones.
>
> > > And this is the part that eluded me when I experimented with this stuff
>
> > > in the 90's: Your parser will encounter tokenized applesoft statements.
>
> > > That means if your code looks like: &HOME then the actual data will be
>
> > > the tokens &(0xAF) and HOME (0x97).  You can use this to your advantage
>
> > > by reusing applesoft tokens for your amp commands instead of having to
>
> > > compare letter after letter (which is the case if the characters do not
>
> > > have the hi-bit set)  Some tokens have the open paren or equal sign as
>
> > > part of the token as well.
>
> > > As for the ins and outs of how to return out of an ampersand command,
>
> > > again it goes back to the applesoft zero-page locations.  I can't
>
> > > remember if you have to clear/set the carry as a status bit like you do
>
> > > in other Apple firmware routines but it wouldn't surprise me if CLC meant
>
> > > there was no error and SEC meant there was.
>
> > > Also, here's a list of all the applesoft tokens as well as my applesoft
>
> > > de-tokenizer routine:
>
> > >http://sourceforge.net/p/java-ace/code/285/tree/jace/src/jace/appleso...
>
> > > I hope this helps.  I can dig up some better information if this is insufficient.
>
> > > -Brendan
>
> > No, there's no automatic error handling--you have to explicitly report any
>
> > errors.
>
> > I replied in the original thread before noticing your new one. ;-)
>
> > Check out the AMPERNADA module in my NadaNet source for a compact, quite
>
> > capable ampersand handler.
>
> > -michael - NadaNet 3.1 and AppleCrate II:http://home.comcast.net/~mjmahon
>
> Top Notch!  Thanks again Michael!!
>
> Did I get the basics right?  I was citing that all from memory and it's been quite a while since I've read about or done anything like that.  I think I was really reciting most of it from what I knew after typing in Nibble listings for some sort of ampersand routine long long ago.
>
> -B

Yes, you got the basics right, except the Ampersand vector is not a
pointer to your routines. It is a JMP statement. So, when Applesoft
encounters the &, it branches to $3F5 and executes whatever is there.
Usually it's a JMP instruction. I suppose you could do something else
in that three byte space but it wouldn't be useful. :P