Path: utzoo!utgpu!utstat!jarvis.csri.toronto.edu!neat.cs.toronto.edu!lamy
Newsgroups: comp.sys.sgi
From: lamy@ai.utoronto.ca (Jean-Francois Lamy)
Subject: Re: Third Party Disks for 4D/25
Message-ID: <89Oct27.225152edt.2687@neat.cs.toronto.edu>
Date: 28 Oct 89 02:52:06 GMT

I have seen several cases where third party hardware was actually making bugs
in the vendor's OS show up more dramatically.  This may have actually saved
the vendor some money and a lot of hair investigating irreproducible bug
reports that where transient when their own equipment was being used (for
example, Revision 14 of the Sun 4/280 CPU board was the first one that could
support both X.25 (synchronous I/O) and stick to the VME specs well enough for
a Ciprico Rimfire board (and no doubt others) to run at its design speed).
Neither of these things were "supported" by Sun or done with Sun software,
but there is no denying that the bugs were there, in the hardware, all along!

I would fully expect a company touting a SCSI driver (which is part of Irix)
to fix it in the event that a bug in the driver gets found, even if none of
the device they sell triggers the bug, or else to be honest enough start
calling the product an "SCSI-like driver" (that would have to happen for
SCSI-like devices too :-).

Part of the selling point for SCSI is the wide availability of useful devices.
Exabytes caused customers to complain about bugs in the vendors' SCSI
software, and in the drives' firmware.  And the vendors who fixed their own
bugs and worked around bugs in hardware their customers wanted got happier
customers, and presumably more sales.  And then some vendors even started
supporting and selling them... 

If I was a vendor and someone I can reasonably trust told me that doing
something reasonable with a reasonably common device used to work and doesn't,
I'd have someone have a look under the hood.  I fully understand that my
definition of reasonable may bear little resemblance to that of a company with
finite resources and trying to turn in a profit for this term.

Jean-Francois Lamy               lamy@ai.utoronto.ca, uunet!ai.utoronto.ca!lamy
AI Group, Department of Computer Science, University of Toronto, Canada M5S 1A4