Re: Results of my straw poll.



In article <SH+ydvNjGpSV@xxxxxxxxxxxxxxxxxxxxxxxx>,
koehler@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx (Bob Koehler) writes:
In article <JFEPKAPBPMDFDBOIANGDOEJEHIAA.dallen@xxxxxxxx>, "Dan Allen" <dallen@xxxxxxxx> writes:

Amen brother! Sections M, U, and P of the docset. I've been around both Unix and
VMS pretty much from the git-go and I've always listed the VMS documentation as
one of VMS's bigger advantages over Unix.

The worst documentation I had to deal with was the HP-UX man page for
the magnetic tape interface. It told me that the minor number of the
special device file contained bit fields to control auto rewind,
density, compression, and such but didn't show the layout of the bit
fields. I had to experiment with a variety of files in /dev to reverse
engineer the controls I needed.

I think the author assumed if I was looking at minor device numbers I
must have the source.

Unless HP-UX became even worse over the years than it was the one time
I was forced to use it, that should not have been an issue.

There are separae device names that allow access to all of the features
supported by a particular tape interface. And that goes back at least
as far as Ultrix-11.
Extracted from the Ultrix-11 man page:
The files /dev/mt? are 800bpi; /dev/ht? are 1600bpi;
/dev/gt? are 6250bpi.
The files /dev/ht?, /dev/tk?, /dev/tm?, and /dev/gt? are
normally rewound when closed, files beginning with the 'n'
prefix are not rewound when closed (for example,
/dev/nrht0).

Looking at FreeBSD this convention is still in use with additional entries
for things like compression and even wether or not to auto eject the tape
after rewind on close.

bill

--
Bill Gunshannon | de-moc-ra-cy (di mok' ra see) n. Three wolves
bill@xxxxxxxxxxxxxxx | and a sheep voting on what's for dinner.
University of Scranton |
Scranton, Pennsylvania | #include <std.disclaimer.h>
.