windows vs linux
Jonathan de Boyne Pollard
RRA> implications on system complexity and/or readability. In order
RRA> to configure something in Linux/Unix (say, a network address),
RRA> you have to traverse a rather traumatic code path: The system
RRA> startup routine reads configuration files and frequently
RRA> interprets these as, say, TCL scripts. These scripts read
RRA> their configuration information from other files, typically
RRA> pass the entries in those files to system utilities which
RRA> reformat the whole thing and pass the configuration information
RRA> via IOCTl calls to the system modules. NT "simply" stores the
RRA> configuration data into the registry, which means by far fewer
RRA> steps from the data to its application.
It is wrong to make generalisations from network address configuration in
Linux and Unix. This is something that varies between different distributions
of Linux (The way that network interface IP addresses are stored in Red Hat
Linux is wildly different to the way that they are stored in SuSE Linux.), let
alone between Linux and Unices. (The configuration files for network
interface IP addresses in Solaris 2.7 are different to both of those Linux
distributions, for example.)
Maxim S. Shatskih
> Linux and Unix. This is something that varies between different distributions
> of Linux (The way that network interface IP addresses are stored in Red Hat
> Linux is wildly different to the way that they are stored in SuSE Linux.), let
> alone between Linux and Unices. (The configuration files for network
> interface IP addresses in Solaris 2.7 are different to both of those Linux
> distributions, for example.)
...and this is a GREAT drawback of UNIX.
Knowing Linux is not knowing UNIX. On any new UNIX, I would need to have a headache of where in /etc the IP addresses are stored.
Max
Peter Köhlmann
I do not see where that drawback should be. The differences in setup are
quite minor compared to how different NT4 is from W2K, let alone XP
config-wise. *Those* are radically different. Headaches are sort of
garanteed. So what is your point?
Peter
--
Get the new Windows XP. Now with eXtra Problems included
Ian Smith
>
> ...and this is a GREAT drawback of UNIX.
> Knowing Linux is not knowing UNIX. On any new UNIX, I would need to have a headache of where in /etc the IP addresses are stored.
Then take some aspirin. If you are incapable of using grep & find just
stay away from Unix full stop.
--
----------------------------------------------------------------------------
Ian Smith
----------------------------------------------------------------------------
Tor Slettnes
Maxim> ...and this is a GREAT drawback of UNIX. Knowing Linux is
Maxim> not knowing UNIX. On any new UNIX, I would need to have a
Maxim> headache of where in /etc the IP addresses are stored.
Will this cause you grief in some way?
How many different flavors of UNIX are you planning to operate, Maxim?
FYI, IP addresses for local lookups are always stored in /etc/hosts,
on any flavor of UNIX or Linux.
(Windows stores that in <windir>\hosts, <windir>\lmhosts,
<windir>\system\hosts, <windir>\system\lmhosts,
<windir>\system32\hosts, or <windir>\system32\lmhosts, depending on
your OS).
There are other nameservice that are used for host lookups also:
NIS, LDAP, DNS, and thanks to Mr. Gates - WINS. Each name service
is configured differently, but generally similar from one OS to
another. Except Windows, of course.
-tor
--
Får i ulveklær
Mark Addinall
>
> RRA> Finally, NTs strategy of system configuration has very big
> RRA> implications on system complexity and/or readability. In order
> RRA> to configure something in Linux/Unix (say, a network address),
> RRA> you have to traverse a rather traumatic code path:
Eh?
vi /etc/hosts
> The system
> RRA> startup routine reads configuration files and frequently
> RRA> interprets these as, say, TCL scripts. These scripts read
> RRA> their configuration information from other files, typically
> RRA> pass the entries in those files to system utilities which
> RRA> reformat the whole thing and pass the configuration information
> RRA> via IOCTl calls to the system modules. NT "simply" stores the
> RRA> configuration data into the registry, which means by far fewer
> RRA> steps from the data to its application.
>
> It is wrong to make generalisations from network address configuration in
> Linux and Unix. This is something that varies between different distributions
> of Linux (The way that network interface IP addresses are stored in Red Hat
> Linux is wildly different to the way that they are stored in SuSE Linux.),
Eh?
> let
> alone between Linux and Unices. (The configuration files for network
> interface IP addresses in Solaris 2.7 are different to both of those Linux
> distributions, for example.)
Eh?
Have either of you actually seem a *NIX box?
--
____ ---------------------
/ /\ addinall engineering
/ / \ mark addinall
/ / \ addi...@tik.com.au
/ / \ 0417 795 327
/ / /\ \ satellite
/ L / / \ E \ wireless
/ E / /\ \ S \ WAN/LAN
/ D / / \ \ C \ project design
/ O / / \ \ H \ project management
/ G / / \ \ E \ embedded systems
/ / / \ \ R \ C programming
/ / / \ \ \ unix administration
/ / / \ \ \ perl
/ / / \ \ \ solaris, linux, bsd,
/ `-------------------\ \ \ sco, aix. hp-ux
/ B A C H \ \ \ TCP/IP. X.25, VPN
'----------------------------' \ \ -------------------
\ \ /
\________________________________\/
Tor Slettnes
>>>>> "Jonathan" == Jonathan de Boyne Pollard wrote:
Mark> Eh?
Jonathan> It is wrong to make generalisations from network address
Jonathan> configuration in Linux and Unix. This is something that
Jonathan> varies between different distributions of Linux (The way
Jonathan> that network interface IP addresses are stored in Red
Jonathan> Hat Linux is wildly different to the way that they are
Jonathan> stored in SuSE Linux.),
Mark> Eh?
Mark> Eh?
Mark> Have either of you actually seem a *NIX box?
Mark,
Unfortunately the articles you responded to have expired from my news
server (that's what I get for being a AT&T sucker^H^H^H^H^H^Hcustomer).
However, the quoting style of Jonathan (Emacs' SuperCite) suggests
that he indeed is a UNIX/Linux user. Not only that, but one that is
deeply enough into UNIX to use Gnus as a newsreader - and one that has
configured it a bit too. That's not something newbies do.
Moreover, from what I could read, he had actual and factual arguments
in his posting, which is much more that the "Eh?"s you provided.
I don't neccessarily agree with the above quoted paragraph, but
nonetheless, if you are going to gain any credibility in their
discussion (and, it seems, for your choice of OS), perhaps some actual
points would be nice.
Joe Potter
Hey guys, aer you fellows referring to the default install of ipv6 under
Suse and ipv4 under Red Hat?????? First time I saw a Suse ifconfig
command I was taken aback.
--
Regards, Joe
Registered Linux User 225822
Man is a bundle of habits, Windows is a bundle of bad habits.
GreyCloud
Looks to me like you found another FUDster.
Mark Addinall
Well, after implementing UNIX on a PDP-11 some years ago,
swinging UCSD P-System from my Pinnicle and Stride 440
in favour of Motorala UNIX for 68000 architectures,
landing into the mess that is (was) Xenix, feeling a lot
better with SCO Unix, At&T 3.2, 3.4, AT&T V.4, BSD,
Discovering SunOS, Then Interactive, then a Solaris suite,
started having a sqiz at Linux in 1994, then the latest and greatest
from AT&T USL, more Linux, SCO Open Server, OSF1 (pig), NeXt,
HP-UX 9.x, 10.x, AIX, just recently Mosix and IBM UNIX for S390,
the way to setup a network has stayed reasonably familiar over
the last two decades. File systems have changed but usually
managed some backward compatibility. SOmetimes a file slips into an
odd place (Sun and HP are notorious, or were, for this)
and the admin tools change from platform to platform <shrug>,
but vi() is always around and the /etc/structure is pretty long
in the tooth. Only scary change I remember was the
new init.d structure being adopted 1990 I think. That
confused me for a few days.
I don't know what 'difficult' version of *NIX these guys are talking
about, but installing Red Hat 7.1 is a matter of sticking the
CD in. Took an hour to get little brothers windoze boxes
on the net, and Samba and masquerading using iptables.
Getting networks to happen as far as the box is concerned
is a doddle under *NIX compared to other platforms.
DOS 1. No directory structure. That was fun heh? Luckily
no one wanted to network. Dos 2, still the same except we
got directories. Developers in software houses still had
one (or more) stand alone boxes on a cigarette strewn desk.
DOS 3 started to hiccup along networking by presenting a few
VERY buggy applications to serial link. Required though as every
different version of "DOS Compatible" had a different bloody file
system and memory manager. So one kept the development
tree on a ridgy-didge IBM and then ported code to other beasts,
NEC, NCR, DEC Rainbow, HP 150, Apricot, Victor, Canon,
Wang, Panasonic. The different 'flavours' of the OS (well,
monitor) addressed memory differently, drive speed, stepping
and densities differed and quite often the interrupt
handlers did different thing. About this time Novell 1.x
was thrust onto us. What a bloody headache. DOS 4 got
released and the world stopped working for two weeks until
the horrendous bugs were quickly fixed. Can't remember
DOS 5, don't think it happened, then DOS 6, about the same as DOS
3 but 20 times as large. Novell were now firmly the choice
for little tiny networks. UNIX was launched for Intel in
the shape of Xenix. Very nearly good UNIX but a trifle
stuffed by Uncle Bills involvement. Quickly dropping
UNIZ for a flash new newcomer, He, along with IBM and
Intel held up and gave unto the world OS/2 v1.13. Finally
an operating system for microcomputers that had some proper
features, HLAPPI and ELHAPPI 3270 data stream,
LU6.2/PU2.1 Advanced Peer to Peer Communications, bi-directional
data tranfers between CICS and OS/2 PM though APPC.C.
Then Bill had a falling out with IBM and foisted a hack they
had been playing with. A bastardised DOS 3 with the OS/2
PM GUI bolted to it. Thus the world begat windoze and windoze
begat windoze 3.11 for workgroups and these little boxes started to
string them selves together with dire warnings "back up every hour",
"if anything goes beep, turn it off for a bit" and other cornerstones
of what we accept as "computer science". Much much later when
the world generally came to regard computers as something that
SHOULD NOT be expected to work, Bill hired a bunch of clever blokes
hurried along by Dave and they sat in a room and re-wrote
VMS. Not a bad job. Yah the world had begat NT.
IBM was floundering with OS/2 but still making big roads into
blue companies and guvmints. They bet the house on SAA,
apart from a few malcreants that went off to dream about AIX.
SCo doing reasonably well with about 80% of all installed
Unixes on the planet. No-one believed in free software,
why would be, we were all making a fucking packet. If you
happened to live on the 'dark side', network and infrastructure
design, you could snaffle a 100Kpa job complete with
Porsche and credit card. Not bad 12 years ago. The free software
people
shouted at Bill at his Wembley conference and Bill laughed back
that they were all mad.
In the background a couple of odd blokes, Steve and Woz, realeased
a whole bunch of things called Apples. They mad a truck load
of money and introduced another architecture that was compatible
with noting, talked to nothing. Curiously after a few
stumbles it seemed to get along with DECNet, whitch also
talked to nothing.
[To present]
My networks generally have inbetween 1000 and 20,000 people
hanging off the wire (or fibre), generally windoze users.
The foolish people (observable on every project) who have fallen
for the guff and stuck critical server systems on NT are
paying the price. Bad performance, no realiability, constant
change of direction from microsloth, and now, a brand new
license fee. My last network (three weeks ago) was for a
1,200 seat network. The bill (sic) to keep the desktops in
operating systems and Word and the rest of the standard stuff
was a scad over a million bucks per year. And the NT servers living
in my cool room usually got bounced every morning 'Just in case'.
The Solaris racks zoomed along forever. The IBM mainframe
was typical 99.9999% uptime.
I have worked on all of the above machines, all of the above
operating systems (including a few not mentioned) and all around the
world, so I think I have a bit of experience. All my macines
are a mixture of Linux and Solaris8. My OS is free, my tools are free,
and my machines stay up until I tell them to go down. I have always
hired
people (some hundred) with the assumtion that if you know one UNIX
well, you pretty much know all of them, and been proved correct
for years and years and years.
And I can still sit in front of damn near anything and
addinall# cat /etc/hosts | more
And I get a little look at how the net is strung together.
So I think my eh? has significance, eh?
Mark.
>
> -tor
>
> --
> Får i ulveklær
--
blah
> And I can still sit in front of damn near anything and
> addinall# cat /etc/hosts | more
UUOC.