Linux-Advocacy Digest #987, Volume #29 Wed, 1 Nov 00 16:13:03 EST
Contents:
Re: Ms employees begging for food (T. Max Devlin)
Re: Ms employees begging for food (Ketil Z Malde)
Re: Ms employees begging for food (Clayton O'Neill)
Re: Ms employees begging for food (Joe Doupnik)
Re: Why Red Hat is as bad as Microsoft (.)
Re: Why don't I use Linux? (Matthias Warkus)
Re: Why don't I use Linux? (Matthias Warkus)
Re: Ms employees begging for food (chris ulrich)
Re: Why Linux is great (2:1)
Re: Why Linux is great (2:1)
----------------------------------------------------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.arch,comp.os.netware.misc
Subject: Re: Ms employees begging for food
Date: Wed, 01 Nov 2000 14:00:18 -0500
Reply-To: [EMAIL PROTECTED]
Said Weevil in comp.os.linux.advocacy;
[...]
>"Sockets," of course, are not required for network communication. It just
>happens to be the abstraction that is used on the most pervasive of networks
>in existence: the internet.
So far as this discussion goes, obviously "TCP/IP internetworking" is
what is being referred to when use the term "the network".
>Communication takes place with or without
>sockets. The physical network people have unrelated layers of communication
>protocols to contend with, and you're right, they needn't concern themselves
>with "sockets," since at their level, sockets are irrelevant. Any sort of
>non-socket protocol can as easily operate on the physical network.
That remains to be seen, though you certainly should be able to make
such a presumption.
>I'm not sure I see what any of this has to do with whether a "socket" can
>exist without someone on the other end.
I honestly cannot understand why.
[...]
>> You don't seem to realize; TCP/IP uses connectionless communications.
>> There is always someone at the other end, theoretically. In reality,
>> there is never "someone on the other end", in a synchronous type of way.
>> Except as an abstraction on the software level of connectivity, where
>> your perspective originates. My premise is not mistaken; despite the
>> desire of system programmers to treat sockets as files, it would be
>> inefficient to do so, because they are not strictly local resources.
>> Frankly, it just isn't for the software to care if there is an "other
>> end", if there is anyone there, or if they are expecting communications;
>> you cannot make assumptions about anything which isn't a local resource.
>
>You seem to be making my point. "Someone on the other end" is not necessary
>for a socket to exist. For that matter, a physical network is not even
>necessary for a socket to exist. A socket is, to strip it to its barest
>essentials, a locally maintained data structure that is used by local
>applications for referencing incoming and outgoing data.
>From your perspective, yes. From anyone else's, no. You say, on the
one hand, that "someone on the other end" is not necessary for a socket
to "exist", but on the other hand you say it is used for referencing
incoming and outgoing data. One must presume that said data comes from
or goes to somewhere, for the abstraction of a socket to be relevant,
correct? So, in fact, the existence of the socket is, in the end,
merely the local abstraction of "someone on the other end". This
abstraction is, of course, still supportable, from your perspective,
whether there is *actually* someone on the other end or not. That is
what makes it a worthwhile abstraction - you can manipulate the socket
the same whether someone is actually on the other end or not. But this
is also the reason why you cannot treat it like a purely local resource.
Yes, there *may* not be someone on the other end, but there *may* be,
and in most useful cases, there is. So to open a socket as you do a
file would be to simply _presume_ that someone is at the other end, and
make no allowances for the possibility that this isn't the case in any
particular instance. But since you have to establish a reference for a
non-local resource (the other end, and data coming from or going to
there) in order to "open" a socket, once that's done, it doesn't matter
if there is another end, let alone whether somebody is there. You can
treat the socket as a file, as all local resources on a Unix box are,
because the network (logical level, in particular in this case, the
transport layer) abstracts the socket as a connection between hosts,
just as you abstract it as a connection between local and remote
resources.
>> Let the network deal with opening sockets, even if you then use them as
>> if they were files (and, again, simply relying on "the network" to
>> handle anything that isn't non-local, like transporting the packets).
>
>The network doesn't "open sockets." Sockets may *use* the network, but the
>network exists, and is used, with or without sockets. Neither needs the
>other in order to exist.
I'm afraid we're just tripping over epistemological semantics. Neither
the socket nor the network "exist", except conceptually. To your
software, the socket *is* the network (where data comes from or goes to,
remember?). The issues are similar to the most common examples of
internetworking: is there a "connection" between browser and web server,
ever? On what level of networking, what layer of abstraction, which OSI
layer, does this connection exist, and is it non-existent, or merely
irrelevant, at any other level/layer?
[...]
>Two programs on the same computer can happily communicate using sockets,
>with or without network connectivity. For that matter, different threads in
>the *same* program can use sockets to communicate, though that would be a
>pretty bad design decision.
Precisely. The wise developers of BSD apparently generalized the
concept of sockets to allow them to support any "ends", not merely the
"this end local; that end remote" presumption which predicated the
development. Or perhaps, unaware to me, they simply used the
pre-existing concept of "sockets" as the way to access virtual
connections established on the transport layer from a particular port on
a particular host to some other particular port on some host.
>TCP/IP is not necessary for sockets to exist. They may use other protocols
>as well. (Not sure about Windows sockets -- they used to only use TCP/IP
>and UDP/IP. This may have changed.)
TCP/IP may not be necessary for the *concept* of sockets to exist. But
the ones we're (or at least I am) referring to are TCP/IP sockets. They
*could* use other protocols, but I'm not interested in thought
experiments. Not practical enough to be worth examining for accuracy or
consistency, as neither is a hallmark of thought experiments.
>A socket does not need another socket, or even a network, to exist. It is
>just a data structure. A local one, at that.
Spoken like a programmer, who thinks that being "just a data structure"
prevents a socket from existing outside a local host. It isn't "another
socket" on the other side, using the concise and comprehensive term
socket. To your perspective, I am sure you'd argue the opposite. But
"the socket", in the abstraction of internetworking, is the virtual
transport-layer connection established between two ports on ostensibly
(but not necessarily) two hosts. "Ports" are hereby defined as
transport/session interface 'addresses', corresponding to individual
client and server processes running on a particular computer, and "host"
being defined as the computer, as identified by one or more IP
addresses.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
======USENET VIRUS=======COPY THE URL BELOW TO YOUR SIG==============
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
====== Posted via Newsfeeds.Com, Uncensored Usenet News ======
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
======= Over 80,000 Newsgroups = 16 Different Servers! ======
------------------------------
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.arch,comp.os.netware.misc
Subject: Re: Ms employees begging for food
From: Ketil Z Malde <[EMAIL PROTECTED]>
Date: Wed, 01 Nov 2000 19:05:28 GMT
T. Max Devlin <[EMAIL PROTECTED]> writes:
> No, true to my contrary nature, I can't agree with you there, either.
You're quite mad, you know that?
> For one thing, I can't agree because I have no regard and no concern for
> what is "simpler for the administrator".
Why not? Complex administration cost money, you can buy a lot of
network hardware for the price of a network engineer.
> Yes, because these particular examples of LAN implementations you
> describe are not at all illustrative of the point I am making.
So you agree that with larger networks, switched is the way to go?
>> Why and when, exactly, is shared better than switched?
>> And why does it have anything to do with utlization statistics?
> Basically, it is better whenever you don't now "why and when, exactly"
> which will be better. :-)
So if I "know" switched is always better...?
-kzm
--
If I haven't seen further, it is by standing in the footprints of giants
------------------------------
From: [EMAIL PROTECTED] (Clayton O'Neill)
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.arch,comp.os.netware.misc
Subject: Re: Ms employees begging for food
Date: 1 Nov 2000 19:47:17 GMT
On Wed, 01 Nov 2000 13:33:20 -0500, T. Max Devlin <[EMAIL PROTECTED]> wrote:
|Said Peter da Silva in comp.os.linux.advocacy;
|>Look, if I have a 16 port shared hub, and it's got 3 stations transmitting
|>at 1 Mbps you're looking at a utilization of what, 30%?
|
|More like, if you have 30% utilization, and you've got 3 stations
|transmitting equally, you will find the throughput of each to be 1 Mbps.
|But that would only hold true in this simplistic model you seem to be
|hung up on, which entirely ignores the reality of the logarithmic
|response curve which Ethernet exhibits. In the real world, if you had 3
|stations transmitting 1 megabit per second on average, your utilization
|would probably be well over 30%.
And see, this is the fundamental disconnect. You believe this, and no one
else here does, probably because it's not true.
------------------------------
From: [EMAIL PROTECTED] (Joe Doupnik)
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.arch,comp.os.netware.misc
Subject: Re: Ms employees begging for food
Date: 1 Nov 00 12:32:33 MDT
>> More like, if you have 30% utilization, and you've got 3 stations
>> transmitting equally, you will find the throughput of each to be 1 Mbps.
>> But that would only hold true in this simplistic model you seem to be
>> hung up on, which entirely ignores the reality of the logarithmic
>> response curve which Ethernet exhibits. In the real world, if you had 3
>> stations transmitting 1 megabit per second on average, your utilization
>> would probably be well over 30%.
==========
In a word, no.
This raving is costing lots of bandwidth. May I humbly suggest running
those experiments to see first hand the true state of affairs. Try several
simultaneous FTP streams amongst a collection of stations in the same
collision domain. The wire is fully occupied, the throughput is rather evenly
divided amongst transmitters, the aggregrate throughput is about 90% of total
wire capacity. Borrow a hub and please give it a try; no model is involved.
While the test runs flip throught the Boggs et al paper on Myths
and Reality concerning Ethernet, location cited previously and a copy is
also on netlab2.usu.edu in directory misc as file ethercap.zip, same file
in pub/mirror/misc on netlab1.usu.edu.
Joe D.
------------------------------
From: [EMAIL PROTECTED] (.)
Subject: Re: Why Red Hat is as bad as Microsoft
Date: 1 Nov 2000 20:23:20 GMT
Roberto Alsina <[EMAIL PROTECTED]> wrote:
> El mi=E9, 01 nov 2000, . escribi=F3:
>>Roberto Alsina <[EMAIL PROTECTED]> wrote:
>>>>> You say SuSE ships a non-patched kernel? What Suse would that be?
>>>>
>>>>You're getting "patched" confused with "entirely re-written".
>>
>>> Well, the original claim was that Mandrake was not Linux because=20
>>
>>> "Its not semantics actually, its legalities. They changed the kernel
>>> without either Cox's or Torvald's approval; therefore it is not=20
>>> linux."
>>
>>> So, choose what you want to say and stick to it, please.
>>
>>I have. You arent understanding the terminology.=20=20
> Maybe.
>>> And I pretty much doubt Mandrake rewrote the kernel, anyway.
>>
>>I never said they did, you moronic fool. I said they rewrote the HEADER=
S.
> Allow me to quote from this very post:
>>>>> You say SuSE ships a non-patched kernel? What Suse would that be?
>>>>You're getting "patched" confused with "entirely re-written".
> Since I had not mentioned headers, and I was replying to something that =
didn't
> say headers, how the fuck is one supposed to guess you actually mean HEA=
DERS,
> when saying KERNEL? If you are too stupid to write coherently, spare the=
world
> your idiocy.
Because I said "headers" in my first post in this thread on the subject. =
I=20
suspected that context would last, but I guess I underestimated you.
> Saying that I am confuding patched with re-written in "a non-patched ker=
nel",
> obviously means you are referring to a rewritten kernel. Unless you are =
not
> using english.
You are having problems with terminology AND context.
>>Thats NOT a patch. Thats a rewrite.
> Whatever.
Well, since you do not care to understand the difference, you do not care
to understand the issue. Since this is the case, I suggest bowing out now=
.
>>>>>>You dont know what youre talking about.
>>>>
>>>>> I may be wrong about slackware (after all, I have not looked at it i=
n 3 years),
>>>>> but Suse, Red hat, Mandrake, Dbian and most others do ship patched k=
ernels.
>>>>
>>>>Again, theres a difference between writing a module to make reiserFS w=
ork and=20
>>>>entirely re-writing the headers because they were "dirty".
>>
>>> What headers did Mandrake rewrite? Example accompanied with version in=
formation
>>> would be nice.
>>
>>Everything is available on their webpage.
> Any more specific pointers? You know, you are saying they did something =
very
> wrong, perhaps something more than your word would give you credibility.
Look around, nitwit. Find a mandrake kernel, find a slackware kernel, do =
a diff
on headers, read the magic.
=====.
------------------------------
From: [EMAIL PROTECTED] (Matthias Warkus)
Subject: Re: Why don't I use Linux?
Date: Tue, 31 Oct 2000 18:50:36 +0100
Reply-To: [EMAIL PROTECTED]
It was the Mon, 30 Oct 2000 21:18:48 GMT...
...and Pete Goodwin <[EMAIL PROTECTED]> wrote:
> [EMAIL PROTECTED] (David Brown) wrote in
> <8tjfd6$v2n$[EMAIL PROTECTED]>:
>
> >"Type" is what you do at the keyboard - why do you think that "type" is
> >somehow more intuitive than "cat"? "ls" is short for "list", "dir" is
> >short for "directory" - neither one is more or less cryptic than the
> >other.
>
> type file?
>
> Seems pretty obvious to me.
Yes. Type is what allows me to type a file. It's Windows' text editor.
...
Hey! Wait a minute!
mawa
--
[...] there was no region where American capital did not support local
labour. Moreover the American press, gramophone, radio, cinematograph
and televisor ceaselessly drenched the planet with American thought.
-- W. Olaf Stapledon, _Last_and_First_Men, 1931
------------------------------
From: [EMAIL PROTECTED] (Matthias Warkus)
Subject: Re: Why don't I use Linux?
Date: Tue, 31 Oct 2000 18:52:31 +0100
Reply-To: [EMAIL PROTECTED]
It was the Mon, 30 Oct 2000 23:53:36 +0000...
...and David Dorward <[EMAIL PROTECTED]> wrote:
> Pete Goodwin wrote:
>
> > dir is directory what is list? List of files? Lists? What? Why 'ls', why
> > not 'list' - why so short?
>
> Probably becuase it takes fewer keystrokes to type.
Yes. The file management command set is all two-letter commands.
cp
dd
mv
ln
ls
rm
Handy, even in these days of tab completion.
mawa
--
Sircar's Corollary to Godwin's Law: If the USENET discussion touches
on homosexuality or Heinlein, Nazis or Hitler are mentioned within
three days.
------------------------------
From: [EMAIL PROTECTED] (chris ulrich)
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.arch,comp.os.netware.misc
Subject: Re: Ms employees begging for food
Date: 1 Nov 2000 20:36:09 GMT
In article <[EMAIL PROTECTED]>,
T. Max Devlin <[EMAIL PROTECTED]> wrote:
>You say you can get much more than 10% or 30% aggregated throughput, but
>the issues is the non-aggregated throughput. It takes a CSMA/CD
>transmission channel (apart from the "point to point"
>thought-experiment) roughly ten times longer to get an arbitrary amount
>of data to the "other end" of the channel when the average utilization
>is at 30% than it does when the utilization is 10%.
>T. Max Devlin
You're simply incorrect.
I've turned off and on whole labs of 50 X terminals at once, while
using other computers for other things. The X terminals all bootp/tftp
their boot images at once (generating lots of traffic) and yet all the
while I still get 400-700k/sec transfers with ftp and interactive response
with telnet/ssh/xterm sessions is only very slightly degraded, if at all.
Since all the X terminals got their boot images (in around twice as long
as it takes for a single one to boot) I'd have to assume that the network
fairly gave bandwidth to all who asked for it, and that it did so with
reasonable efficiency. Judging from the collisions on the network, I'd
say that it was at nearly 90% utilization. And this is on that same messy
math network I described in a prior post.
If you absolutely must have five nodes on a network who all must
250k/sec of bandwidth, or even 200k/sec of bandwidth, I'd hesitate to
stick them all on the same ethernet, but if those five need only 190k/sec,
I wouldn't hesitate. 10 users at 98k/sec? Sure. They all will get it
day in and day out for years. Because ethernet works, is fair and efficent.
Now, if I had 9 people who each need 50k/sec and one who needs 500k/sec
and no way of limiting those other 9 to only 50k/sec, then I would worry,
but that is a senseless thought experiments that doesn't reflect how things
really work (that 500k/sec would just get a different network).
You may indeed sell lots of products by saying that you must subscribe
10 times the bandwidth you expect to use, especially if your typical
audience is a regulation sized PHB. There is a reason dogbert the evil
consultant typically avoids posting his sales pitch to usenet news;
perhaps you should follow his example.
On the plus side, if you convince someone to buy ten times as much
crap as they really need, chances are you will probably solve their
problem or at least not cause new ones. Hardly good engineering, though.
chris
------------------------------
From: 2:1 <[EMAIL PROTECTED]>
Subject: Re: Why Linux is great
Date: Wed, 01 Nov 2000 22:42:13 +0000
George Richard Russell wrote:
>
> In article <[EMAIL PROTECTED]>, Aaron Ginn wrote:
> >[EMAIL PROTECTED] (George Richard Russell) writes:
> >
> >> All hidden behind a backwards UI for 70's teletypes.
> >>
> >> Shame that.
> >>
> >> George Russell
> >
> >You haven't used KDE2 yet, have you?
>
> Yup.
>
> Its got a nice *terminal emulator* included.
Real text terminals are better (esp 180x80 on a 21' moniter).
> Even supports colour console applications.
Are you touting this as a god or bad thing?
> Comes with a mini cli panel applet and handle cli short cuts
> for man , info , web, help, smb, nfs urls etc.
That's good...
> Has a terminal that embeds in the file manager
Good too...
> Yet it's still Unix commands that you type.
Still good...
> Wow.
Nor *really* that suprising considering it's a UNIX workalike...
> For the retro look, choose the Green on Black console schema.
You can get white on black VT220s. I was using one today, in fact.
> KDE, however nice, is limited in scope, and no Unix desktop
> will ever shed its cli roots
I should hope not. The command line is extermely powerful. It is much
easier to do complex commands in the CLI than on a GUI.
ls *.mp3 | sort | sed -e "$(cat -n tracks | sed -e's/\([0-9][0-9]*\)[
\ ]\(.*\)/\1s!\\(\.\*\\)!mv \\1 \"\2\.mp3\"!/')" | sh
That's a small script I happen to use a lot. What's the easiest way of
doing it in a GUI?
(btw, it takes a bunck of mp3s names audio_01.mp3 etc, fresh from a
ripper and renames the first to the first line in a file called tracks,
the second to the second etc etc very useful for ripped CDs.)
This isn't a convoluted example, it is something I use regularly and it
sves me quite a bit of drudgery. How can a GUI do the same easily.
It's reasons like that that I don't want to loose the UNIX cli.
-Ed
--
Konrad Zuse should recognised. He built the first | Edward Rosten
binary digital computer (Z1, with floating point) the | Engineer
first general purpose computer (the Z3) and the first | u98ejr@
commercial one (Z4). | eng.ox.ac.uk
------------------------------
From: 2:1 <[EMAIL PROTECTED]>
Subject: Re: Why Linux is great
Date: Wed, 01 Nov 2000 22:44:29 +0000
George Richard Russell wrote:
>
> In article <[EMAIL PROTECTED]>, Aaron Ginn wrote:
> >> Its got a nice *terminal emulator* included.
> >So does Mac OS X. You don't have to use it if you don't want to.
>
> That may be true for Macos X, but its not, and likely never will
> be, true for Linux.
>
> Try and setup Linux and use it without going to a console.
>
> >> For the retro look, choose the Green on Black console schema.
> >>
> >> KDE, however nice, is limited in scope, and no Unix desktop
> >> will ever shed its cli roots
> >
> >Nor should it have to. Contrary to popular belief, the CLI is still
> >alive and well, and a better choice than a GUI in many cases.
>
> Every few years, Unix gets another GUI. Its a shame the cli isn't
> replaced / improved as often.
The CLI is made of two things: the shell and utility programs. There are
loads of different shells avaliable, and there are plenty of utilities
avaliable too. There are probably many being created as we speak.
Besides, the CLI is very functional as it is.
-Ed
> >Win9x still hasn't shed its DOS roots by the same argument.
> >
> >REGEDIT?
> >MSCONFIG?
>
> Two commands that appear in no version of DOS, your point is?
>
> FWIW, they were *new* in Windows 95 - based on DOS - but your
> point in no way shows this.
>
> George Russell
--
Konrad Zuse should recognised. He built the first | Edward Rosten
binary digital computer (Z1, with floating point) the | Engineer
first general purpose computer (the Z3) and the first | u98ejr@
commercial one (Z4). | eng.ox.ac.uk
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.advocacy) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Advocacy Digest
******************************
