HP9000 info/review - (nf)

[utzoo!decvax!harpo!seismo!hao!hplabs!neil]

#R:rocks34:-17200:hplabs:12700002:000:2144
hplabs!neil    May  7 01:03:00 1983


I saw some comments about the HP9000 Unix workstation, and wanted
to correct some factual errors.  Please note that while I work
for HP, this is not "offical" informations (i.e. I do not work
for that divisions marketing department...).

The letter was in net.unix-wizards from rocks34!dw, Don Wegeng,
Xerox Corp.,  Rochester, NY, (716) 422-3347

    In terms of the Unix implementation, it is suppose to be derived from
    v7, with some of the Berkeley utilities (and maybe some System 3, I'm
    not sure).  This version seemed to have most of the utilities that I
    remembered to look for, including vi, lint, more, uucp, news, etc.
    Adb was missing, and I don't know if they plan to include it in the
    release version.  Otherwise, I saw everything that I wanted.

The system was derived from system III, with some utilities ported
from Berkeley, plus several HP-local applications and utilities.

    My only gripe with the system is that HP has joined the group of Unix
    suppliers which has decided to *improve* the user interface by making
    it consistant, at the expense of compatability with other versions.
    For example, if you want ps with a long listing you can't type
    "ps l", you HAVE to type "ps -l".  They have also changed some of the 
    flags to ps.  I personally find this to be VERY annoying. 

An interesting comment.  The 9000's unix is not a port.  Instead it
is a layering of the Unix system interface on top of an already
existing real time kernel.  There is no /dev/kmem, and PS was rewritten
from scratch.  The flags were changed because PS is very dependent
on the underlying implementation.  The "ps -l" is NOT a bug -- read
the system III manual (it annoys the hell out of me too, but...).

All in all the 9000 is impresively compatible with Bell system III
for a) a layered system and b) a different machine architecture.
Despite a hackers mentality of "improving" everything, the rules
were "if it ain't broke, don't fix it." We put in about five man
years in testing the kernel for "Unix-hood", and management gave
much importance to compatibility.


    Neil Katin
    ucbvax!hplabs!neil