Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!yale!think.com!hsdndev!cmcl2!panix!alexis
From: alexis@panix.uucp (Alexis Rosen)
Newsgroups: comp.unix.aux
Subject: Re: A/UX 2.0.1 questions
Message-ID: <1991Mar17.075747.1576@panix.uucp>
Date: 17 Mar 91 07:57:47 GMT
References: <50005@apple.Apple.COM> <1991Mar10.182907.9845@panix.uucp> <50137@apple.Apple.COM>
Distribution: na
Organization: PANIX - Public Access Unix Systems of NY
Lines: 119

In article <50137@apple.Apple.COM> ksand@Apple.COM (Kent Sandvik) writes:
> alexis@panix.uucp (Alexis Rosen) writes:
>>>Well, the trick with patching _HFSDispatch in order to fake multiple
>>>HFS volumes on one single volume has always been a hack, and thus
>>>is not suppored by neither A/UX or MacOS.
>>
>>Please. It works. Thousands of people use it, and you should support it.
>>No debate is permissible on this subject. :-)
>
>Well, MacOS emulation specs are based on the Inside Macintosh documentation,
>and there's never been a official policy of supporting patches that are 
>not defined by Apple documentation. This is also true of System 7, which
>may or may not break INITs that patch _HFSDispatch.

Sorry, my meaning wasn't clear. I don't know what you're talking about with
this _HFSDispatch business (but I'm guessing it was some poor way of doing
multiple Mac partitions on a disk). I was referring to the totally legitimate
way that several formatters, such as the MicroNet, put multiple MacOS volumes
on a simgle disk. They follow the partition map specs in IM-V exactly.

A/UX's lack of full support is a flaw in A/UX, and as far as I can tell,
unjustifiable from a cost-to-implement standpoint (when you scan the partition
map at boot time for a MacOS partition, just scan the whole map- don't stop
when you get the first one... this is not a lot of code...)

BTW, I'll add that the MicroNet formatter is probably the best in the business.
They've had bug-free support for A/UX partitions for over two years. And,
unlike most drivers (including Apple's), theirs has been 32-bit clean for a
long time, and thus didn't have a problem of crashing under System 7 in 32-bit
mode.

>>>Otherwise HD Setup with Apple harddisks asks for a rich set of
>>>possible A/UX setups before the installation. The installation
>>>software is quite different compared with the old A/UX 1.1
>>>installation program.
>>
>>This is the real problem. It is disgraceful that HD Setup won't partition
>>3rd party drives. I wouldn't mind if it could put drivers on them too, but
>>that's not so critical. But Apple refuses to sell reasonably large disks,
>>and sticks us with the archaic and arcane dp. I was under the strong
>>impression that this was going to change in 2.0.1, but if I did hear such
>>a commitment, it wasn't kept. What happened to "easy to use" and "great user
>>interface"?
>
>A couple of days ago I understood why we won't support TPV drives
>with HD Setup. I had this problem with a LaCie drive, where using
>the Silverlining partition program cured the hard disk. Because 
>the Silverlining program *programmed* the Hard disk not to send
>obscure SCSI command calls that the generic A/UX dev. driver
>filters out. 
>
>The reason for filtering is simple - the UNIX kernel should not die
>suddenly [while MacOS allows this, I hope this will be fixed soon
>under MacOS as well].

You're implying that LaCie can write formatters better than any Apple
programmers. This is certainly historically true, but there's no reason
it has to stay this way.

>Anyway, if a customer reads that HD Setup can partition *any* 
>TPV hard disk, and tries with HD Setup, which can't possibly know
>all the possible firmware programming setups and firmware control
>codes for every TPV hard disk drive, then....
>
>In the worst case he gets angry, and if he's rich and is really mad
>he will sue Apple - and court cases are not fun.

This is a straw man. You don't have to _guarantee_ anything. But you _could_
list the drives it did work with- if you tested with drives from Seagate,
Quantum, Maxtor, Conners, and Miniscribe, you'd make 95%+ of your potential
users happy. And as you probably know, the above-named vendors' drives are
quite compatible with A/UX. It wouldn't take much (if any) work to teach HD
Setup about them.

>I hope that each one sees this dilemma. The simple case would be
>to define a generic call, something like "A/UX sanity reprogramming
>control code", and HD Setup sends this call to the TPV hard disk
>when the end user wants to use the hard disk for A/UX.
>
>Then again, we can't predict that all the Third Party Vendors will
>implement this scheme, and we are back in the twilight zone of 
>angry customers and liability issues.

It is not their responsibility to establish a SCSI 1.1 standard with special
Apple commands. It is *your* responsibility, as a vendor trying to sell a
product into a crowded marketplace, to make your product as convenient as
is possible.

>>This whole thing makes me angry because Apple has already done the work.
>>All it would take to make HD Setup work would be the deletion of a few lines
>>of code.
>
>If life was that simple.... SIAC and Apple is able to test those drives
>that Apple sells so they work with A/UX partitions without problems.
>
>Anyway, this area is very important for the future, and hopefully
>by providing and assisting TPV hard disk vendors with information
>we could avoid the problem with non-working hard disks and working
>partition programs.

This is simply not sufficient. As long as you take this attitude you (not
personally, Kent- "you" means Apple) have failed to provide a complete
and sufficient solution. You ignore the fact that many users, *especially
large corporate and goverment users*, will not buy third party stuff, or
would rather avoid it at all costs. The less dependance on it, the better.
(Of course, that also brings up the question- When will Apple start selling
real disk drives instead of the 80MB toys? {And that inflammatory question
is exactly the question- and the tone- you'll hear from many buyers!} )

A few days ago I bought a Seagate/Imprimis Elite 1300MB disk. (And it is
*FAST!*) I hooked it up to an RS/6000, and with SMIT I had it up and running
in minutes. The Elite didn't even exist when this IBM system was built.

If IBM can do this, so can you.

---
Alexis Rosen
Owner/Sysadmin, PANIX Public Access Unix, NY
{cmcl2,apple}!panix!alexis