Dude,

calm down - we're all friends here (I hope :-).

I'll address your only reasonable question to me - the one that didn't call
me or my ideas stupid and deals with applications that sit atop the
operating system.

Sorry to shock you all, but I am a (part-time) system administrator and
(full-time) security architect who sees general purpose operating systems as
the necessary HAL that allows me to run FrobbleBits, where FrobbleBits is
any application that provides a business driver to allow my customers to
make money, lose less, or be competitive. Typically this is a firewall,
database, middleware (transaction servers), a word processor or a billing
system. Nearly all of these end user {services,daemons,apps} needs to be up
as soon as the boring bit (the OS) is up. To do that, I'd like the OS to
start the app, and shut it down cleanly when I shut down the box for
whatever reason. rc.d does this cleanly. I was a dba of a health rdbms that
looked after 1 million patients (about 1/3rd of the state I used to live in)
as well as a system that HAD to be up or people would die. Before I moved to
Sydney, I helped setup and administrate 54 distributed SQL servers,
associated HA cluster and the middleware needed to count the votes at a
state election. This system looked after 4-5 million transactions in a six
hour period as well as performing several million transactions to organise
elections in record time (3 days worth of CPU time versus 20 days of 20+
hour days for staff). I now design and secure country-sized advanced IP
networks and services and rejig telcos and scare networking vendors. I think
I'm perfectly qualified to have a view, equally as valid as yours. My views
are large scale. My projects each deal in the hundreds to thousands of
heterogenous systems. NetBSD is not large scale, but it _could_ be. Being
different - you have to be damn fine to be different or you're not in.

FrobbleBits is what my users and I personally am interested in. The
underlying OS is of almost zero interest to me. I use, administrate and hack
on a lot of them. I choose to spend my *spare* time hacking on NetBSD
because I want to. But when I get to work, if Solaris is required, I'll do
Solaris. If Win2K does it for me, I'll implement Win2K. Ditto with Linux.
The trick is to say "NetBSD will do it for me" so it becomes a valid option.
Attitudes like this that basically prevent app developers or system admins
from deploying on NetBSD prevent NetBSD from being a valid choice. Even
FreeBSD is less of a chance than NetBSD at where I work (a major telco).

I came to hack on NetBSD out of preference to Linux because there is CVS
control of the kernel and essential bits and it has a working kernel
debugger. The NetBSD kernel has a nice MI bus architecture and my Alpha just
worked. Trust me, I could go back to using solely Linux at home, but I chose
NetBSD because my best friend Luke Mewburn develops on it, and I felt it
offered me technical advantages. Those advantages came about because NetBSD
*progressed*. I'm not into using System III or BSD 4.2. I want a modern
operating system to hack on.

NetBSD's current base world is a sort of frozen snapshot of what was useful
to most people in the late 80's/early 90's. For example, why include bind
and sendmail, but not gopher, or some form of web server? Sure, they're
available in pkgsrc, but sendmail is not even the best MTA out there anymore
(flames to /dev/null - you know I'm right and Exchange 5.5 MTA will soon be
everywhere that counts :-)

There's nothing wrong with progress. Good progress, that fulfils users (both
current and potential) needs, whilst being technically demonstrably better
than other solutions will always meet with my approval (and hopefully many
more besides just me).

thanks,

Andrew van der Stock, ajv@greebo.net http://www.greebo.net
SAGE-AU President	                   http://www.sage-au.org.au

-----Original Message-----
From: port-alpha-owner@netbsd.org [mailto:port-alpha-owner@netbsd.org]On
Behalf Of Lord Isildur
Sent: Tuesday, 28 November 2000 7:26 PM
To: Andrew van der Stock
Cc: port-alpha@netbsd.org
Subject: RE: removing packages

[snip]
yes, because i and _many_ others advocate AGAINST this ridiculous talk of
packagifying BSD!

[snip]
it's stupid _and_ unnecessary.

[snip]
linux box is because you have NO CLUE otherwise what's there. on a BSD
system, you know EXACTLY what's there- there is one unified distribution.

[snip]
this is also stupid. [...] If you don't like BSD, use linux, but
don't clamor to make BSD into crap!

[snip]
you mean databases in what sense? these have nothing to do with the
operating system. If youre talking aout postgres,mysql,informix,etc,
those each have their own administration issues. this might be something
behind why so many people hire database administrators, just an idea.
databases are no more a part of the operating system than a word processor
or an image viewer. I suppose you'd like all database vendors/makers to
standardize on their administrative techniques, too?

[snip]
yes, everything should be EXACTLY THE SAME. Bow down to the ONE TRUE WAY.
This sounds like you want a micro$oft system! go use their 'only one way
to do it' systems!

[snip]
be the vastly superior UNIX, for a multitude of reasons. I am _extremely_
sickened by people pressuring BSD to be like linux just because of
popularity or market trends/whims!