basic App Key support in BASIC? Need insight.

[tschak909]

Hello, everyone. 

 

@jamm has checked in the bits of code to support what we call "App Keys" on the #FujiNet.

 

More documentation info can be found here: 

https://github.com/FujiNetWIFI/fujinet-platformio/wiki/SIO-Commands-for-Device-ID-%2470  as commands $DB to $DE

 

These provide a key/value store for applications to store and retrieve configuration information on the FujiNet.

 

Since these are implemented as SIO calls, and are outside of the purvey of the N: device, they need to be accessed as direct SIO calls, so a solution needs to be made that can be easily embeddable in BASIC.

 

The idea is to embed machine code in strings, something like:

 

OPENKEY$="..."
CLOSEKEY$="..."
READKEY$="..."
WRITEKEY$="..."

For Open/close, you'd have use a 6 byte string, e.g. KEY$, containing the app key to open/close.

For Read/Write, you'd have a 66 byte output, VAL$, containing the 64 byte contents to return.

 

How can we make it so that we can do:

 

KEY$="......":X=USR(ADR(OPENKEY$))
X=USR(ADR(READKEY$)):? VAL$
VAL="New Value":X=USR(ADR(WRITEKEY$)

 

Thoughts, guys?

-Thom

[+x=usr(1536)]

What about using the AUX2 field in the OPEN / CLOSE / INPUT / PUT / GET, etc. commands to specify a memory address for the routine in question, then embedding the requisite assembly string between the ending quotes?  Something like:

 

OPEN #1,12,1536,"N:assembly-as-atascii-string-goes-here"

 

Then use the regular Atari BASIC I/O commands to receive output from and interact with the resident app?

[tschak909]

not possible. aux2 is an 8 bit value, and there is no meaningful way to use the devicespec in the way you propose.

 

-Thom

 

[+x=usr(1536)]

Yeah, my fault - I meant to specify aux2 as a memory page value, not the decimal equivalent.  Derp.

 

My thought was that the handler could intercept the aux2 value, poke the supplied string in at that page number (not necessarily ideal, but at least representable in 8 bits), and then call it.  But if there's no way for the devicespec to know that it's being sent assembly, it's a moot point.