Accessing partitions in drive images

[Ben Scott]

On Sun, Jan 29, 2012 at 4:45 PM, Joshua Judson Rosen
<rozzin at geekspace.com> wrote:
>> Flash drives wouldn't work as floppies, so they're treated as fixed
>> disks.
>
> What does this mean? Why wouldn't USB sticks or MMC/SD cards work
> `as floppies'?

  Because then we wouldn't be able to use a flash drive/card bigger
than 2.88 MB.

  (I think 2.88 MB is the biggest floppy disk size defined by IBM-PC
conventions.  But if there are others, they're around that size.)

  This isn't about terminology or how the user wants to use the flash
drive.  It's about how these things fit into the scary world of the
IBM-PC.  Thirty years of hardware, software, firmware, and interfaces
cobbled together by countless vendors.  Crufty old BIOS code, written
in assembler by people long gone and patched endlessly.  Several
different OSes (DOS, classic Windows, Win NT, several *nix variants),
each with their own software stack to handle disks, partitions, media
types, and so on.  Hundreds of low-level utilities.  Every single
thing with its own assumptions about how storage works.

  For anything to fit into that mess, it has to be shoe-horned into
the original design assumptions.

> Also..., what does the "fixed" in "fixed disk" mean?

  "Fixed" as in fixed-in-place.  Non-removable.

  Of course, USB flash drives *are* removable, but there's nothing in
the IBM-PC design (circa 1985) that allows for a third option.  It's
either a floppy or a fixed disk.  It wouldn't work as a floppy, so
it's a fixed disk.

> I do have a vague recollection of the SD/MMC architecture as... something
> that plugs directly into the host bus and contains its own IDE controller?

  I actually don't think that's right, but it doesn't really matter
for current discussion.  What matters is getting these things to work
in an old and eclectic IBM-PC ecosystem.

> And I know that USB drives identify themselves as `general mass storage'
> devices (generally...) ...

  That's the USB device class.  That's fine if the only thing involved
is USB stuff.  But when you want to use it with an IBM-PC, you're
adopting tons of historical baggage.  The device has to be *presented*
and *treated* in a way that's compatible.  So while the USB controller
can handle all sorts of things, the OS/BIOS/utilities/etc. are more
limited.

> The thing that's actually piqued my interest is that I recently received
> a whole-device (partitionless) FAT-formatted USB flash-stick... from
> a Mac user.

  Apple has strict control over their computing ecosystem.  While that
tends to annoy a lot of techie types, it does mean that they can
introduce new technologies easier, since they're the ones in charge.
Anything that doesn't do it their way is defined to be wrong (even if
it's one of their older products that's now declared "wrong").

>>   Next up: Why are console windows traditionally 80 columns wide?
>
> Because 80 is an optimal mix of `nice round radix-2 number' and
> `nice round radix-10' number?

  Because console windows/xterms are based on old-school dumb
terminals, which were traditionally 80 columns wide.  And *they* were
80 columns wide because the most common IBM punched  (paper) card was
80 columns wide.  And those cards are the size they are because they
were designed to fit into bins made for old US currency, circa late
1800s.

  If you're thinking that seems like a rather haphazard chain of
events, you're right.  Welcome to IT.  :-)

 -- Ben