[comp.sys.amiga.tech] Tape Backup

jdepp@deimos.ADS.COM (Joe Depp) (11/29/89)

Where, indeed, are the tape backup units for the Amiga?  Doesn't anybody
else get tired of backing up their hard disk to floppies?  Has Commodore made
it so difficult to implement that no third party has the courage to try?

Please interpret this, not as flames, but as a cry of anguish!

lphillips@lpami.wimsey.bc.ca (Larry Phillips) (11/29/89)

In <9890@zodiac.ADS.COM>, jdepp@deimos.ADS.COM (Joe Depp) writes:
>Where, indeed, are the tape backup units for the Amiga?  Doesn't anybody
>else get tired of backing up their hard disk to floppies?  Has Commodore made
>it so difficult to implement that no third party has the courage to try?

I recently purchased a Comspec host adapter, so I could do some work with the
SCSI driver and SCSI tape on my other controller.  I knew it was slow, but the
price was right for a temporary way to get my data out of the way of errant
driver code. I noticed an interesting thing about the Comspec. It uses two
drivers.  I assume (from a cursory disassembly), that one handles the low level
hardware stuff, and the other handles the actual device stuff on the Amiga end
of things, and that the higher level one gives commands to the lowert level
one. This led me to wonder if a tape device could be added as another driver.

A call to Comspec resulted in confirmation that it could be done, but that they
had not does it yet. They were also planning on releasing some pogrammer
information for those who wanted to roll their own, but had not yet gotten to
it.

The saddest part was that they had decided not to write the tape stuff because
"tape is too expesive and there is little or no market for it". (Hey.. their
feelings, not mine).

They did say they would let me know, either directly or through a posting to
the net, but so far nothing, and its been a while.

Meantime, I have a backup working, and a restore about half done. Nothing
sophisticated, but it's fairly fast (backup of 36 megs takes about 19 min.),
and it will do the job for me. Unfortunately, I will not be able to release
what I do have, until I can confirm that it will work on the HardFrame (on its
way to me, via the mail, so I should get it this fiscal year :-)

Meanwhile, I have some questions for the net.techies....

My backup and restore programs use SCSIDirect, er... directly. This works quite
well, and allows me to send any commands to any device, and to get status back
from them. In thinking about the Comspec driver, I had the idea that a 'generic
tape driver' (or any other non-disk device driver) could be written to interact
with the actual device driver, as long as that driver used the SCSIDirect
standard.

The way I envision it working would be that it would look like any 'normal'
driver,  but would actually translate tape specific commands into SCSIDirect IO
calls, and send them to the actual driver for the host adapter. Later, a
'TapeFileSyetem' might be used to allow things like 'copy foo ST0:'.

First question.. does anyone see anything inherently wrong with this idea?

Now, in doing direct SCSIDirect, as I am doing now, I simply ignore any return
code in D0 that comes back from the IO, and check on the status by looking ta
the scsi_status byte, doing a sense if anything is wrong, to get more info. In
speaking with a friend who is also doing work on a tape backup/restore program,
we seem to have a disagreement as to what constitutes an error return He feels
that the error return from a DoIO should be related to a hard error, such as
DMA being hosed up or the drive not being able to process the command at all. I
feel that it should indicate _any_ abnormal condition that neds further
investigation. In particular, I feel that 'tape not loaded' should be indicated
in both io_error and the return code in D0, while he feels that it should be
considered a 'normal' condition.

Second question... what should D0 contain after an DoIO if there is no tape in
place?

-larry

--
" All I ask of my body is that it carry around my head."
         - Thomas Alva Edison -
+-----------------------------------------------------------------------+ 
|   //   Larry Phillips                                                 |
| \X/    lphillips@lpami.wimsey.bc.ca -or- uunet!van-bc!lpami!lphillips |
|        COMPUSERVE: 76703,4322  -or-  76703.4322@compuserve.com        |
+-----------------------------------------------------------------------+

stephen@hpdml93.HP.COM (Stephen Holmstead) (11/30/89)

FelineGrace@cup.portal.com (Dana B Bourgeois) writes:
>I know it probably isn't a problem now but when will support for more
>than 8 SCSI devices appear?  And do the current products support the
>attach/detach elements that are built into the standard?  SCSI has 
>been around for at least 5 years now and it seems there is only support 
>for a subset of the standard.  What's the reason?

Cost.  As one who writes embedded SCSI controllers for a living, I can
say that when we create a SCSI device, we go through the SCSI spec and
say, "yes, people use this feature" and "nobody uses this feature, so we
won't put it in".  The more SCSI commands that we have to support, the
higher the price tag that we are going to have to put on the device.  In
fact, we support TONS of the SCSI commands that are OPTIONAL according
to the SCSI specification.  We can't afford to spend money to support
commands that controllers don't use.  It doesn't make sense.

BTW, SCSI-2 is on the scene.  Lots of new features.  Check it out!

>And why don't we have SCSI modems?  Don't tell me a 9600 BPS modem
>wouldn't benefit from SCSI.  Better yet, why wouldn't it be a good idea?

Probably because of cost.  But I forsee more devices becoming SCSI.  I
was looking through the SCSI spec and there are commands like "PLAY
AUDIO", "READ CD-ROM CAPACITY", "START PRINT", etc.  In fact, there are
commands to support communications devices (modems), printers, scanners,
processor devices, tape drives, and even automatic medium-changing
devices!  I guess if there was a demand, then people would start
supporting it.  But currently, to support SCSI on an IBM PC requires
purchasing a $400 emulex controller.  Thus, SCSI doesn't have much
saturation on PCs, and thus, most hardware manufactures don't make
modems, printers, etc, for SCSI.  When the PC world starts to support
SCSI, then you will probably see more SCSI peripherals.  If you were a
modem-maker, why would you make a modem that costs more than the ones on
the market and can't be used by the bulk of the computers on the market?
Sad, but true. :-(

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Stephen Holmstead
...!hplabs!hpdmlge!stephen                      //
Hewlett Packard Disk Memory Division          \X/  Amiga Forever!

"I remind you that humans are only a tiny minority in this galaxy."
	-- Spock, "The Apple," stardate 3715.6.