Programmer 7500/MAX_PROGRAMMERS.iso/INFO/MISC/TNH_PC.ZIP/WHITE.NL

Text File | 1987-01-14 | 18.0 KB | 582 lines

An Introduction to PC Diskettes

             Howard White
        Boston Computer Society
          IBM PC Users Group

Introduction

When you buy double-density
soft-sector disks for your PC, or
choose between DISKCOPY and COPY, you
are making decisions affected by the
relationship between DOS, your PC's
Disk Operating System, and the type
of disk drives installed in your
computer. While it is not necessary
to have a technical understanding of
the way DOS maintains information on
your disks, it's useful to understand
the general principles so that you'll
be a better informed user.

This article will introduce you to
"tracks" and "sectors", explain in
disk terms what a file is, describe
the common DOS utilities for
formatting, copying, and testing
disks, and explain the concepts of
density.

Looking at Diskettes

We start our exploration of diskettes
by looking at one. With the permanent
label in the upper left hand corner,
we see:

o   The large round hole in the
    center. When you close the disk
    drive's "door", the drive clamps
    onto the disk here. Many disks
    come with "hub ring
    reinforcements", thin plastic
    rings like those used to
    reinforce three-ring notebook
    paper, intended to help the disk
    withstand the mechanical forces
    of sudden starts and stops.

o   On the right side just below the
    center of the hub hole, is a
    smaller round hole called the
    "index hole". If you very
    carefully turn the disk within
    its protective jacket, you will
    see that at one one point is a
    small hole in the disk hole in
    the disk itself visible.

o   Below the hub hole is a slot
    shaped rather like a long
    race-track, through which you can
    see the disk surface itself. It
    is through this slot that the
    disk drive heads read and write
    information to the disk surface.

o   At the right side, about one inch
    from the top, is a rectangular
    punch from the side of the disk
    cover. This is the "write-protect
    slot". It may not be present on
    all disks, particularly those you
    have purchased with programs on
    them. If the slot is absent, or
    if it is covered with tape, the
    drive can read data from the disk
    but not write on it.

o   Two very small oval notches at
    the bottom of the disk, flanking
    the head slot.  The drive uses
    these notches to assist in
    keeping the diskette in position.

o   As you have almost certainly been
    advised, disks can be damaged or
    destroyed by:

    o   Touching the recording
        surface (though the head
        hole) with your fingers or
        anything else;
    o   Writing on a disk label with
        a ball-point pen or pencil.
        Use a felt-tip marker;
    o   Bending the disk;
    o   Spilling coffee, etc. on the
        disk;
    o   Overheating a disk (e.g. by
        leaving it in the hot sun in
        your car);
    o   Exposing your disk to stray
        magnetic fields.

Information is read from and placed
on your disk by one or two "heads"
(depending on whether you have single
or double-sided drives). Heads used
with "hard disks",  including the
"Winchester" drives available for
PC's, "float" above the disk surface
on a microscopic cushion of air. But
the mini-floppy drive heads actually
touch the disk surface. Thus both the
disks themselves and the heads should
be kept clean - head cleaning kits
are available - so that precious
information will not be scraped off
the disk surface by abrasion.

Tracks and Sectors

From the point of view of DOS, data
on your PC diskettes is organized in
"tracks" and "sectors".

Tracks are narrow concentric circles
going around the disk. Sectors are
pie-shaped slices of the disk.  DOS
1.0 and DOS 1.1 read and write 40
tracks (numbered 0..39) and eight
sectors (numbered 1..8) on disks. If
you have double-sided drives, DOS 1.1
automatically formats and deals with
disks containing  information on both
sides of a disk. The distance between
tracks - and therefore, the number of
tracks on a disk - is a built-in
mechanical and electronic function of
the drive.

When you buy new disks, they are like
blank sheets of paper, containing no
information at all. Formatting the
disk puts onto it the information DOS
needs to maintain a directory and
files. Normally, you also put onto
the disk certain parts of DOS (using
the /S option after the FORMAT
command).

The track nearest the outside edge of
a disk - track 0 - is almost entirely
reserved by DOS for its purposes.
Track 0, Sector 1, contains the "boot
record" which the system sometimes
needs to begin operation. Sectors 2
and 3 contain the "file allocation
table", or FAT, which is the disk
"reservation clerk", keeping records
of which portions of the disk have
information and which have empty
space. Finally, sectors 4-7 contain
the directory - information about the
files, a part of which you see when
you type the "DIR" or "FILES"
command.

In addition, assuming you have
formatted the disk using the /S
(system) option, two "hidden" files
called IBMBIO.COM and IBMDOS.COM are
placed on the disk (in different
location, depending on whether you
have single or double sided drives).
They are called "hidden" because the
normal directory command will not
reveal their presence on a disk.

In computer jargon, all of this is
"transparent to the user", meaning
you don't have to (and generally
can't) decide where information will
go on your disks. But just because
it's "transparent" doesn't
necessarily mean you shouldn't be
aware of the decisions DOS makes for
you.

PC DOS 1.1 puts 512 bytes of
information on each sector of a disk.
As a broad generality, a byte is
equal to a character, but BASIC
normally saves programs in a
compacted format, so you can't make
an automatic judgement about the
amount of disk space a BASIC program
will use.

But for purposes of illustration,
let's assume you have a
newly-formatted disk in your drive
and have created a BASIC program of
2000 characters, or 2K bytes.  You
type SAVE "FILE1", and DOS begins
putting your program onto disk.

On the outermost track, sectors 1..7
already are taken up with system
stuff, as mentioned previously, so
the first available sector is track
0, sector 8. The first 512 bytes of
your program go there. Then, because
DOS always tries to keep things
together, the remaining roughly 1500
bytes go onto track 1, sectors 1, 2
and 3.

Now, you write another, entirely
different, 2000-byte BASIC program
and SAVE  it as FILE2. Again, DOS
puts it on the first available spots,
so it goes to track 1, sectors 4, 5,
6 and 7.

Going back to your first program, add
a few hundred characters, and SAVE it
again under the same filename, FILE1.
The first four sectors (even if
changed) go back to the same spots as
the old first four sectors. But now
your second file is in the way of
keeping things orderly. The next
available free sector is track 1,
sector 8, so that's where the rest of
your revised first file goes.

Next, you decide to delete your
second program, FILE2, using DELETE
or ERASE commands in DOS or the KILL
command in BASIC. These commands
don't actually erase any information
from the disk (therefore making it
possible, using a utility program ,
to "unerase" a program you've just
accidentally wiped out). What the
DELETE family of commands does is
tell DOS that the sectors formerly
occupied by, say, FILE2.BAS, are now
available for recycling.

Now you create and save a new
program, FILE.BAS, which takes up
five sectors.  The first four go into
the spots vacated by the now obsolete
FILE2.BAS. The last one has to go
into the first available spot, after
the newly-added part of FILE1.BAS. We
now begin to have a patchwork quilt.

We're still not in bad shape. But
suppose our disk gets fuller. We're
deleting old files to make room, and
creating vacant areas defined solely
by the needs of DOS at the time the
disk was younger.  It shouldn't be
hard to see that a full disk can
become a real crazy quilt, with any
sector of any file occupying any of
the 320 (single sided) or 640 (double
sided) sectors not otherwise used by
DOS, system files, other user files.

Since all of this is, as we said,
transparent to the user, what
difference does it make? Not very
much, if all you are doing with disks
is storing programs. But as soon as
you start using disk data in your
programs, you can affect the
performance of those programs,
perhaps substantially, by proper disk
management.

To understand why, consider what
happens when you ask for disk access,
either reading or writing. First, the
disk is told to start rotating at 300
rpm - it only spins when needed, and
the red light goes on to tell you
that's happening.  Next, DOS figures
out which track the data is to be
read from or written to. A "stepper
motor" moves the heads toward or away
from the center of the disk,
positioning them exactly over the
right track (if you have double sided
drives, the top and bottom head move
together). This movement of heads
from track to track is the slowest
operation your disk drives performs,
and disk access, in general, is the
slowest thing your computer does
(excluding things governed by
external forces like slow printers
and telephone lines). Once the heads
are over the right track, access to
the desired sector is relatively fast
- the drive "sees" the index hole
open up as that one hole punched in
the disk passes over the openings in
the disk jacket, and then the
drive/controller combination just has
to wait the right amount of time
before reading or writing to a
sector.

The message here is that moving from
one track to another is the slowest
of the slow things a disk drive does.
DOS tries to keep sectors of a
program together, but as the
patchwork quilt pattern develops, it
can't help much. In an extreme case,
the heads may read a sector on track
0, then pick up another from track
39, then move back to track 1 to pick
up a third, then to track 39 again...
and so forth. The slowest part of
disk access is in control.

What can be done?  COPY.

The DOS manual lists two different
ways of copying information from one
disk to another: COPY and DISKCOPY.

DISKCOPY makes an exact copy of one
disk to another. So if one disk is a
patchwork quilt, a DISKCOPY of it
will look the same. But there's an
alternative:

COPY *.*

In this case, asterisks are
"wildcards", meaning "whatever would
legally be here". So *.* means any
legal filename followed by any legal
filename extension. Therefore, if you
type COPY A: *.* B: , you direct DOS
to copy all files on drive A to the
disk on B. There are some assumptions
here. First, the "target" drive - in
this case, B, must have enough room
for the files copied to it. Second,
unlike more sophisticated schemes on
larger computers, DOS presently
provides no protection scheme. So if
there is a FILE1.BAS on A and a
FILE1.BAS on B, COPY A version. COPY
*.* is best done TO a disk whose
contents are expendable.

The advantage, however, is what
happens on COPYing. If the target
disk is clean (containing nothing but
what FORMAT /S put there), COPY *.*
to this disk will reorganize data.
Thus if FILE1.BAS on the source drive
was scattered from track 39, sector
4, to track 3, sector 2, with random
stops in between, after a copy it
will be placed in consecutive sector
order on the new disk. The result is
that the next time you try to read
FILE1.BAS, the heads won't have to
jump all over the place looking for
pieces of the file, and should
therefore be able to read it faster.

The message is: generally COPY x
procedure IF the target disk, y in
this case, is clean. If it is not
clean, COPY will work around the
files on the disk, putting the files
from x in the best possible
positions. DISKCOPY x: y: will make
disk y an exact clone of the x,
wiping out any files that may have
been on y and reproducing the
patchwork quilt of x.

There are utility programs in DOS to
check the accuracy of copying.
DISKCOMP is analogous to DISKCOPY;
that is, it starts with the
assumption that the compared disk is
supposed to be a snapshot of the
original, and reports differences.
Even if each file on the second disk
is an exact copy of the corresponding
file on the first, DISKCOMP will
report differences if they're in
different sector/track positions.

On the other hand, COMP compares the
accuracy of the copy of individual
files. You can say COMP A:FILE1.BAS
B:FILE1.BAS and get a report as to
whether the two files (in this case,
on different disks) are identical.
But you cannot say COMP A B on B:).
The reason for this, of course, is
vagueness - you can't compare
something-or-other here with
something-or-other there and get
(except by chance) any valid
information.

Let us now consider some practical
implications for PC owners of what
has been covered in this article.

Diskette Types

There are three separate choices
offered to the purchaser of 5 1/4"
diskettes (aside from the questions
of quality, price, etc.).

Soft-sectored or hard-sectored: For
the PC, soft-sectored disks have only
one index hole on the disk surface
itself. Once per revolution this hole
is visible through the hole in the
protective jacket, and
drive/controller/DOS uses this hole
to establish the location of the
first sector on any track.
Hard-sectored disks have a hole for
each sector, and thus each hole marks
the beginning of a new sector. If you
tried to use a hard-sectored  disk in
a PC, the machine would get very
confused. Sometimes hard-sectored
disks are not specifically labelled
as such, but specify "10 sectors" or
"16 sectors". Don't buy them.

Single-sided or double-sided.  At
first blush, this would seem to be an
obvious choice. If you have
single-sided drives, you buy
single-sided disks, and if you have
double-sided drives, you buy
double-sided disks. Not so easy. If
you have double-sided drives, you
can, of course, make the extra
investment in double-sided disks, but
you probably don't need to. It is
economically impractical for disk
manufacturers to make some disks with
recording surfaces on one side and
other disks with recording surfaces
on both sides. Today's "single-sided"
diskettes look, and usually behave,
exactly the same as double-sided
diskettes.

The result of this is that -
depending on the brand of disks you
buy - you can generally successfully
format and use "single-sided" disks
in double-sided drives, at
substantial savings in disk costs.
There is no real danger in this
practice. When you format a disk, one
of the things DOS does is determine
whether there are bad sectors (ones
which can't reliably hold
information). If it finds any, it
tells you "xxx bytes in bad sectors"
and it seals them off against any
future attempts to put data there.
The net result is that you have a
double-sided disk with slightly less
than full storage capacity at the
price of a single-sided disk.

Recommendation: Unless you have money
to burn, buy single-sided disks even
if you have double-sided drives. If a
particular brand seems to produce
"bad sectors", consider switching
brands.

Density (single, double, quad).
"Density", in simplest terms, is a
measure of the amount of information
that can reliably be packed into a
given area of a recording surface.
The keyword here is "reliably", for
as in the question of single vs
double sided, the matter may be
settled by how well sample disks
measure up to manufacturing tests.
However, the question of density also
raises the question of how many
tracks are written on your disks.

The drive or drives that came with
your PC - whether single or double
sided, are almost certainly "double
density" drives, also referred to as
"48TPI" (tracks per inch) drives.
This means that when the disk
controller tells the drive "move one
track closer to the center", the
drive will respond by moving the
heads exactly 1/48" closer to the
center. It is possible to buy "single
density" and "quad density" drives
which look exactly like the drives in
your PC. When a single-density drive
is told to move one track closer to
the center, its heads move 1/24:.
When a quad-density drive is given
the same instruction, it moves the
heads only 1/96". The result is that
a quad-density drive puts two tracks
in the same amount of space a
double-density drive uses for one
track, and so the disk holds twice as
much information. There is, at least
potentially, a tradeoff here. As you
squeeze more and more information
onto a 5 1/4" disk, you place
increasing demands on the ability of
the drive to locate information
accurately, and on the ability of the
drive to locate information with the
increasing precision required.

I started this discussion by
mentioning density as one of the
factors involved in buying diskettes.
As with single- vs. double-sided,
manufacturing quality control is a
major part of what distinguishes one
density from another.  Generally, you
will buy double-density drives for
your PC.

DOS 2.0 and DOS 2.1 have the ability
to increase disk capacity by 12 1/2
per cent by changing the formatting
scheme. Thus double-sided disks,
which now hold 320K bytes, will hold
360K without any other changes in
software or hardware.

This is done by increasing the number
of sectors per track from 8 to 9. In
turn, it can do so because PC disks
are soft-sectored, their sector
organization is determined by the
operating system, not the number of
holes punched in the disk.  In doing
so, DOS 2.0 is increasing the density
of disk storage, putting more
information into each sector than
earlier versions of DOS. (Note that
it is not increasing density in the
way quad-density drives do - by
squeezing tracks closer together.) It
takes more than a DOS change to do
that; the disk drives themselves must
be configured for the changed track
pattern.