[Coco] Preserving old CoCo diskettes...
Gene Heskett
gene.heskett at verizon.net
Tue May 18 23:08:05 EDT 2010
On Tuesday 18 May 2010, Roger Taylor wrote:
>At 03:49 PM 5/18/2010, you wrote:
>>On Tuesday 18 May 2010, Roger Taylor wrote:
>> >At 02:04 PM 5/18/2010, you wrote:
>> >>Roger, why not contribute that code to the nitros9 project?
>> >
>> >I didn't write an rz/sz protocol that could work at 115200 or 230400
>> >bps. My point was that if you can't get those old programs to work,
>> >move on to a new idea using the least number of chip signals as
>> >possible. Hence, a software solution.
>>
>>rzsz is a separate problem, which is why I mentioned other protocols as
>> being faster, much faster.
>>
>>Since I compiled the rzsz you are, or have played with, I can't argue that
>>its architecturally wrong, it is, and that means you are limited to a hair
>>over 700 cps on a 6309 machine, and the middle 500's on a 6809 machine.
>>It is, in the face of poor line quality, absolutely bulletproof though.
>> It just keeps on hammering until the checksums do match. The problem is
>> that it builds the checksum one &^% char at a time, so that whole loop
>> has to execute for every incoming byte. Grrr.
>
>Exactly. Always build from a stable program to a faster program,
>with stability coming first.
Well, if one wanted to make some real changes to Chucks code, not that
subbing a table lookup function for the crc calcs wasn't major, but it was
good for about a 200cps speedup over Chucks raw code, what it needs is a pair
of buffers in rz, so it can dump 128 bytes to buffer A, then feed buffer A to
the table driven crc, while reloading buffer B. That would I should think,
be good for several hundred cps of a speedup as the table driven crc lookup,
when handed a page to check, did that at about 10 kilobytes/sec when I had
it carved out as a separate bit of test code 15 years ago.
>> >I've found it a much better experience to just mount remote .dsk
>> >images and have the files sitting right there in native DOS or OS-9
>> >format already good to run or copy.
>>
>>That sounds like a better idea, but I'm not sure I know how.
>
>But then I admit it could take the fun out of watching rz/sz do it's
>thing, or actually writing the protocol and proving you're still the
>mean coder you used to be. ;)
But I have to get into the mindset that makes me absolutely driven, and that
leads to complaints from the better half's bleacher seats. Even I don't
enjoy that anymore.
There might be more than one way to skin this cat, but I really think there
may be a greater chance of starting from scratch, dissing that aciapak driver
from 15+ years ago & figuring out what the diff is between the null modem
cable I used then, and the one I have now. I have NDI what happened to that
original cable I used so successfully when it was working against the serial
port of an amiga 2000.
If anyone has a null modem cable with db25's on both ends that is known to
work without the eventual lockup, having a ring list of that cable would be
at least as good as sliced bread or bottled beer to me.
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
<NullC> I like the seed code for computing masking curves.
<NullC> I've never seen code that made be want to drink before that...
More information about the Coco
mailing list