NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/sys/apple2/19836

Text File | 1992-09-04 | 2.6 KB | 49 lines

Newsgroups: comp.sys.apple2
Path: sparky!uunet!munnari.oz.au!mel.dit.csiro.au!mineng.dmpe.CSIRO.AU!dmssyd.syd.dms.CSIRO.AU!metro!ipso!fawlty!johnmac
From: [e-mail] (John MacLean)
Subject: Re: Apple II RWTS codes.
Reply-To: [e-mail] (John MacLean)
Organization: Tower Technology, Sydney
Date: Fri, 4 Sep 1992 00:43:21 GMT
Message-ID: <[e-mail]>
References: <philip.714873752@labtam> <[e-mail]> <philip.714956159@labtam>
Lines: 37

In article <philip.714956159@labtam> [e-mail] (Philip Stephens) writes:
>>I was under the impression that the Locksmith fast 16 sector copier didn't do
>>any denibblizing of the raw disk data...
>  In fact, it does!  I disassembled the code once, and was quite amazed to see
>that it performed both read and write conversions on the fly.  I don't remember
>the full details now, but when reading it converts the 6-and-2 encoding back
>into 256 bytes of data that is still partially encoded, but which can then be
>converted back into 6-and-2 encoding on the fly.  Pretty interesting stuff.
>  Philip Stephens

I disassembled this too at one stage.
What it means is that it is possible to encode/decode 6-and-2 on the fly -
but not the standard 6-and-2 that DOS and ProDOS use - this cannot be done
on the fly (where on the fly means as you read/write data to disk).
Unfortunately no-one ever released a DOS which used this - or it could have
been even faster than Pronto/Diversi/etc.
It should be possible to have no interleave in real situation if you
cobined this optimisation with the file manager change.

The only code I have seen that uses a read/write like this is some of
Roland G's 18 sector routines (eg: Airheart, etc), and some of his 6
sector ($300 bytes each) routines (eg: Wings of Fury).
However, since they pack so much on a track, you couldn't possibly have
a real file system sitting on top, as the simple overheads of calling
a RWTS routine would mean you would miss the next block (assuming 1-1
interleave).
What he did was to have calls to his RWTS that let you read/write N
blocks starting at block A on track T.
Very neat indeed.
If you want to look at some of this code, good luck boot-tracing, there
are about four levels of EORing and other nasty tricks in the code.
However, if you want to learn more, its worth it.
--
This net: [e-mail]                   Phone: [phone]
That net: uunet!fawlty.towers.oz.au!johnmac             Fax:   [phone]
Snail:    Tower Technology, 1 Apollo Pl,                Home:  [phone]
          Lane Cove, NSW 2066, Australia.