Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!princeton!caip!think!nike!ucbcad!ucbvax!hplabs!sdcrdcf!dennisg
From: dennisg@sdcrdcf.UUCP (Dennis Griesser)
Newsgroups: net.micro.mac
Subject: Re: 800k drive formats (A proposal)
Message-ID: <2849@sdcrdcf.UUCP>
Date: Thu, 19-Jun-86 16:13:54 EDT
Article-I.D.: sdcrdcf.2849
Posted: Thu Jun 19 16:13:54 1986
Date-Received: Sat, 21-Jun-86 10:33:47 EDT
References: <262@sivax.UUCP>
Reply-To: dennisg@sdcrdcf.UUCP (Dennis Griesser)
Distribution: net
Organization: System Development Corporation R&D, Santa Monica
Lines: 62

In article <262@sivax.UUCP> phil@sivax.UUCP (Phil Hunt) writes:
>AS I see it, the 800k format (or double-sided format, if you prefer), is set
>up for optimum head switching,  ie the first x tracks are addressed by head 0
>then the next x tracks are addressed by head 1 and so on throughout the disk.

I heard that all the sectors on one track of side 0 are used and then all
the sectors of the corresponding track on side 1.  This type of allocation
is called "cylinder".

>Why not just allocat sectors 0-800 (approx) to the first side, the 801-1600
>to the second???   It would solve ALOT of current problems (some caused by
>user cheapness, but what the heck)

Allocating data sequentially across an entire side means that you have to move
(step) the head twice as often to get to the data.  And head stepping can take
quite awhile.  That's why Apple did it this way.  It is one of the primary
reasons why the 800K drives are so much faster.

>1) A disk could be initted as 800k, then a copy-protected disk could be copied
>to it in bit copy mode.  A utility could be then written to tweak it back to
>800k and voila, a useable 800k disk with the 400k copy protection intact!!! 
>
>This would work since sectors 0-800 would still be accessed the same as on a 
>400k disk....

Cute, but there should be a better way to do it.  Besides, copy protection is
a bucket of slime anyway.

>2) Users who use Single-sided disks in double sided mode would find that their
>data is MUCH more intact from the initial copying to the 800k disk since the
>directory sectors would ALWAYS be on the first side.  Even if the back of the
>disk went south one day, the initially loaded and most of the files stored on
>the back of the disk could be recovered.  (I just had a disk sector 18, track
>0, head 1 (back of disk) go bad.  That is the 2nd block in the directory, so
>the disk was rendered unuseable).

A poor argument.  Lost data is lost data.  You simply propose losing a different
part of the disk.  What if the front side went south one day?  You lose it
all anyway.

>3) current 400k disks could be turned into 800k disks very easily by a
>utility that could be written to format the back and tweak like in (1) above
>the disk to 800k.  It would save alot of copying off of a disk, format, copying
>back to the disk files to cvt a disk to 800k.

Nice.  There should be a way to do this without scrapping the cylinder
allocation.

. . . . . . . Alternate proposal

In the beginning of each disk are some bytes that tell what disk format is on
the disk.  I believe that these are called "signature bytes".

One value means "400K" one meafs "000C"&  T`ere are fdavgrs fgr EFS afd @FS$ afd sgee fgr Dasa&@gw abgut sgee sagfature bqtes t`at saq "088K(b}|(no|(allocated in cylinders".
And let the file system do the right thing from there.

That lets you get the perceived benefits of linear allocation and pay the
necessary cost, or use cylinders and forgo the "800K is 400K plus some"
similarity.