Down to 5 lines, and they all fit inside the 40 columns without
wrapping, and it's a slightly better program by not having that IF test
on every one of the 1400 bytes, and added one for model 200, and a
slightly larger generic version that works on both 100 and 200 and any
binary that fits inside 2 blocks (2k)
...by just using precomputed address values for the binary. I realized,
why not? It's already full of other brittleness anyway that only works
if everything is exactly as expected. This way it allows the code to
contain the right CLEAR statement so you don't have to manually fix it
after the install, and so don't have to include a prompt just to display
the start address.
And it's only supposed to be like an emergency fallback method anyway.
You'd only need it if it was 1986 and your machine crashes while you're
away from your cassette deck at home or in your car. So you whip out
that printed hardcopy sheet from your briefcase or folded up and stuffed
in the option rom compartment, type it in and recover all your stuff.
It's fine if it's exactly mated to the specific binary that's on the RAMPAC.
Also fixed an off-by-one error in the previous version. It was adding
one extra byte of junk on the end. It worked and ran ok because the
start address was right, and the entire program was included, it just
also had an extra byte of junk on the end that was never touched. Where
the previous version has E=S+L, should have been S+L-1, end = start +
length - 1.
This I now consider tolerable for it's use-case.
1 CLEAR0,61558:S=61558:E=62957
2 OUT129,2:FORJ=0TO15:K=INP(131):NEXT
3 J=S:K=S+1007:GOSUB5:OUT129,1:J=K+1
4 K=E:GOSUB5:SAVEM"RAMDSK",S,E,S:END
5 FORA=JTOK:POKEA,INP(131):NEXT:RETURN
That's the 100 version, the 200 version is the same just with different
values in line 1.
For any other binary, I still have a slightly larger generic version in
there also that does read the .CO header and prompts the user to
manually do the final CLEAR to set HIMEM for whatever address the binary
had. And it works on both 100 and 200 without changing anything.
1 CLEAR0,59000 :OUT129,2 :FORA=0TO9 :N=INP(131) :NEXT
2 GOSUB6 :S=N :GOSUB6 :L=N :GOSUB6 :X=N
3 N=S+1007 :E=S+L-1 :FORA=STOE :?".";
4 POKEA,INP(131) :IF A=N THEN OUT129,1
5 NEXT :?"type CLEAR 0,"S":NEW" :SAVEM"RAMDSK",S,E,X:END
6 N=INP(131) :N=N+INP(131)*256 :RETURN
Still has the IF in the loop instead of making the poke loop into a
gosub just because this way comes out shorter to type, and that really
is the more important feature in this special case.
On 12/16/23 19:07, Brian Brindle wrote:
Nice!
On Sat, Dec 16, 2023 at 5:33 PM Brian K. White <[email protected]
<mailto:[email protected]>> wrote:
__
down to 6 lines
inserted spaces for easier manual read & type
got rid of I variable name to avoid looking like 1
|1 CLEAR 0,61000 :OUT 129,2 :FOR N=0 TO 9 :B=INP(131) :NEXT 2 GOSUB
6 :S=N :GOSUB 6 :L=N :E=S+L :GOSUB 6 :X=N 3 N=S+1007 :FOR A=S TO E
:B=INP(131) :POKE A,B 4 IF A=N THEN OUT 129,1 5 ?"."; :NEXT :?"type
CLEAR 0,"S":NEW" :SAVEM "RAMDSK",S,E,X 6 N=INP(131)
:N=N+INP(131)*256 :RETURN|
And line 5 can actually be reduced to just:
|5 NEXT :CA||L||L X|
--
bkw
On 12/16/23 12:57, Brian K. White wrote:
I now have a working manual BASIC bootstrap that can load
RAMDSK.CO <http://RAMDSK.CO> from the device after a cold restart.
https://github.com/bkw777/NODE_DATAPAC/blob/main/software/RAMDSK/RBOOT.DO
<https://github.com/bkw777/NODE_DATAPAC/blob/main/software/RAMDSK/RBOOT.DO>
It is only 8 lines like the old docs say BOOT.BA <http://BOOT.BA> was
Well, it's 7 lines not counting a comment and a message, or 9
lines counting both.
But they are not very neat or convenient. It's still kind of a lot
to manually type in for a bootstrap, but if you had a print out
you could type it in a couple minutes and then you have RAMDSK
back a few seconds after that.
This is not exactly a convenient way to install, but it should
exist as an option just on principle.
You're on the road, suffer a cold reset, everything's safely
backed up on the RAMPAC, but you need to get RAMDSK installed
somehow.
I was dreaming it might be possible to write something short
enough to print right on the card in the silkscreen. hahaha no. :)
Maybe it can be improved over time.
I also wrote a simple inspector that can read raw data from
anywhere on the devivce, in either bank, and display in either
ascii or hex, and the ascii also shows all the non-printing
control characters as their CTRL-x character in inverse video so
the display doesn't get messed up and no bytes are hidden either.
So 0x00 shows as inverse "@" for instance.
https://github.com/bkw777/NODE_DATAPAC/tree/main/software/RPI
<https://github.com/bkw777/NODE_DATAPAC/tree/main/software/RPI>
The main point of it is to be small and scrutable, uses no machine
code, supports banks.
--
bkw
--
bkw
