On Sun, 26 Nov 2000, Luke Mewburn wrote:
> 
> I find it interesting that you consider the ability to easily manipulate
> a single service (e.g., dhcpd) without resorting to using ps and kill,

ooh without resorting to ps and kill, oh my, how terrible! we have simple 
orthogonal mechanisms which are perfectly adequate, but you can't rest 
easy with that, you need totally crufty mechanisms that are specific to 
_each_ single bloody "service"  (it's a DAEMON folks, not a 'service'. a 
daemon is a _process_, not some mysitcal 'service' wankery!) instead. 
sure, then, you can just memorize the specifics for each 'service' and be 
fine. Just like windows. memorie what to click for thing A or thing B and
you're fine. No need to know anythign else. instead, though, when a single
(albeit requiring some _thinking_!) mechanism exists which can take care 
of _all_ such possibilities (that is, the regular commands ), which you 
might have to use in an intelligent manner, you freak out and run away 
hiding, complaining that it's too hard? i dont think so. that's bogus. 

> and then invoking the service the same way rc did, ``dumbing BSD down
> to appeal to moron wannabe-administrators''. I'm sure a lot of people
> who are far from being morons or wannabes would take umbrage at your
> statement.

let them! 

> 
> With respect to ``the BSD way'', I refer you to the post by Robert Elz
> about how things were often done by ``BSD'':
> 	http://mail-index.netbsd.org/tech-userlevel/2000/03/17/0010.html
> 
> BSD is not about being a nochangenik. In fact, wasn't the reason that

No, BSD is about being the superior UNIX distribution. Part of that is 
that we have developed mechanisms to do things which are simple, 
consistent, and still the UNIX way of doing things: lots of simple, 
general purpose tools that one can (intelligently) combine to get 
anything you want done. The sysV people and linux have gone the other 
direction, making more and more specific commands, which increasingly 
become environments in themselves, and the major reason is either market 
pressure to dumb things down (sysV) or the vast majority of the 
development community being totally ignorant of things being done in a 
much better and more completely thought-out way already, or being total 
amateurs. The vast majority of the development in the linux community 
that i see _is_ being done by complete amateurs who aren't even _aware_ 
that the problems they are finding kludgy and gross (yet accepted as 
'cutting edge' by other more clueless folk) solutions to were solved in a 
much more elegant and orthogonal way fifteen or twenty years before, 
largely by either bell labs folks or CSRG folks. Now that popularity is 
on the side of these kludgy solutions, because theyre mainstream and 
accepted, our superior way of doing things is under pressure to change to 
be more like the popular guys. Well, if NetBSD becomes just another 
popular and _crappy_ unix wannabee, i'll just keep the last sources i 
deem good and fork my own development tree, it will be a lot of extra 
work (i run BSD on three different architectures) but at least things 
will be able to stay superior. I dont think anybody holds a candle to the 
BSD style of distribution, rc scripts, and so on. I think that the guys 
at CSRG (and the great deities at belllabs before them) really hit on the 
_right_ way, it took a long time of development, sure, but once you get 
to the best, you KEEP it there you don't keep adding crap. Ongoing 
development should be NOT in changing the organization and rcs and so on, 
it should be in bug fixes, security fixes, occasional code cleanups, 
improvements in algorithms and correctness, inprovements in size 
efficiency, and so on. maintenance. The period of great architects is 
certainly over, and it's to us to make the small refinements over time 
that keep it shining brightly. 

keep in mind the quote: 'those who fail to understand UNIX will be doomed 
to reimplement it - poorly.' Weve already got the right implementation, 
dont give in to the pressure from those who fail to understand it and 
keep on making poor imitation implementations!

> the original Berkeley Software Distribution started was because people
> wanted extra and/or different stuff than the stuff shipped by USL?

It was because the folks at UCB made improvements to the kernel, first, 
and only later did other things come in, and eventually it become a whole
UNIX distribution. 

Isildur