Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!apple!ksand@apple.com
From: ksand@apple.com (Kent Sandvik)
Newsgroups: comp.unix.aux
Subject: Re: non-apple disks
Message-ID: <13971@goofy.Apple.COM>
Date: 8 Jun 91 17:41:41 GMT
References: <1991Jun5.123635.4268@am.dsir.govt.nz> <1991Jun5.141103.23420@helios.physics.utoronto.ca> <13902@goofy.Apple.COM> <1991Jun7.100146.20572@helios.physics.utoronto.ca>
Sender: usenet@Apple.COM
Organization: Apple
Lines: 36

In article <1991Jun7.100146.20572@helios.physics.utoronto.ca>, sysmark@aurora.physics.utoronto.ca (Mark Bartelt) writes:
> [ Kent Sandvik ]
> 
> | Well, here we go again. Let me give another example of how hard it is to support
> | third party vendor SCSI disks. The general assumption is that a hard disk should
> | start with asynchonous mode, and start a form of handshaking talking with the other
> | end if they want synchronous mode, and at what speeds. 
> |
> | Well, there are hard disks out there that maybe or possibly starts immediately in
> | synchronous mode, and wonders why the Mac does not talk with them.
> 
> A disk that behaves that way is broken.  Unless I'm mistaken (if so, please
> correct me), the async/sync negotiation is part of the SCSI *standard*, not
> just a "general assumption".  A drive that starts out in synchronous mode
> without getting an OK from the other end isn't following the standard, and
> its manufacturer should be loudly screamed at.  I don't think any rational
> customer would fault Apple for not being able to deal with a drive which
> violates the SCSI spec!

The problem is the definition of a 'rational customer' - I don't think there
are many customers that know the SCSI specs inside out.
 
> |                                                                    If we said that 
> | HD Setup would work with *any* hard disk, and a customer gets into trouble to a similar
> | case, then we are liable.
> 
> We don't expect you to promise to support *any* hard disk.  At least, not
> disks which don't conform to the SCSI spec.  But it would be nice if you
> *did* support well-behaved drives.

This is the problem, it is hard to define the line where the 'well-behaved SCSI disk
set' starts and the bad guy SCSI disks ends. Anyway, I'm afraid this will lead
to another round of SCSI hard disk A/UX bashing, so I will stop talking about this
thing :-).

Kent