Denys misspeaks:
> Actually, MPE is pretty basic when it comes to tape handling.  It is nowhere
> near as sophisticated as UNIX or NT/2000.

Hmph.

Not in my book.

Starting with the ability to programmatically specify density,
to the ability to detect retries, and the ability to get *real*
I/O error information, and ending with ease of use and
quality of operator interface, I'd have to rate
MPE well ahead of Unix and NT.

However, it isn't worth arguing about opinions, so I won't.

I've done some research on generic tape copying utilities on
Unix, including talking with the authors of some.  So far,
the only tape copying utilities Ive found are rather simplistic.
(e.g., the TAPECOPY model).  (Not surprising, I guess.)

We ported our copying utility, but are unhappy about some
missing Unix functionality: ability to detect retries on reads
(without that, you *CANNOT* know how reliable your input tape
reading was); and ability to have multiple outstanding read/write
requests.  (The AIO facility on the machines we've tried seems to
be weakly implemented at best, and then only for disk access.)

--
Stan Sieler
           [log in to unmask]
www.allegro.com/sieler/wanted/index.html
       www.allegro.com/sieler

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *