Linux-Advocacy Digest #221, Volume #26 Sat, 22 Apr 00 17:13:09 EDT
Contents:
Re: which OS is best? (mlw)
Re: Which Linux and which hardcopy? ("ismet")
Re: which OS is best? ([EMAIL PROTECTED])
Re: on installing software on linux. a worst broken system. (Joe Kiser)
Re: which OS is best? (Mig Mig)
Re: which OS is best? (mlw)
Re: Sell Me On Linux (Christopher Browne)
Re: Grasping perspective... (was Re: Forget buying drestin UNIX...) (SeaDragon)
----------------------------------------------------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.ms-windows.advocacy,comp.sys.mac.advocacy,comp.os.ms-windows.nt.advocacy,alt.flame.macintosh
Subject: Re: which OS is best?
Date: Sat, 22 Apr 2000 16:25:04 -0400
David Corn wrote:
>
> mlw wrote:
> >
> > David Corn wrote:
> > >
> > > Craig Kelley wrote:
> > > >
> > > > [EMAIL PROTECTED] writes:
> > > >
> > > > > On 21 Apr 2000 18:40:45 -0500, [EMAIL PROTECTED] (Leslie Mikesell)
> > > > > wrote:
> > > > >
> > > > > >>Did I say domain controller? No - I said NFS and SMB sharing. And
> > > > > >>for sharing in WinXX, click the device to be shared, click SHARING...,
> > > > > >>and away you go. It's far simpler than Linux.
> > > > > >
> > > > > >OK, try again. Which button did you click on NT to get tha
> > > > > >NFS sharing done and how long did it take to find?
> > > > >
> > > > > NT doesn't come with it. The objective is to get OS-native sharing
> > > > > going - somehow, anyhow, with a minimum of fuss. NT (and 95/98) do
> > > > > that very well. Linux doesn't. Editing /etc/exports for NFS, for
> > > > > example, isn't my idea of fun. ksysv and such make the automation of
> > > > > such things easier, but I don't consider it 'easy' by any stretch.
> > > >
> > > > So use Linuxconf. You can't complain about the lack of tools which
> > > > exist.
> > > >
> > > > [snip more oh-my-god-I-have-to-use-a-text-editor stuff]
> > >
> > > The point, which is correct and perfectly valid, is that Linux is far
> > > more difficult to set up for even basic filesharing.
> >
> > There is no evidence that supports that pointing and clicking through
> > several dialogs is any easier to use than a text editor. No scientific
> > study what so ever. While one can have an opinion, one can not say for
> > sure. And BTW, there is a huge difference between easy to use and easy
> > to learn. Something slightly harder to learn but much easier to use
> > (text files) may, in fact, be better than something that is slightly
> > easier to learn, but consistently harder to use (levels of dialog
> > boxes).
> >
>
> I just can't buy it. I can tell someone in about 3 seconds how to get
> this going in Win98; can you do the same in Linux (SMB sharing vs NFS
> sharing)?
>
> Bear in mind:
> The person on the other end probably can't type.
> The person .. can easily make a typo.
> " " doesn't know anything about IP addresses.
> " " doesn't know if he's using DNS or not, so machinename resolution
> can't be trusted.
> " " may or may not have a NIS server.
>
> For an office setup, sure, with NIS it isn't that bad once you've done
> it a few times, but for one-offs, *I* wouldn't want to be the one to set
> up someone else's filesharing over the phone.
This is the point of my argument, easier to use is NOT easier to learn.
A person, sufficiently practiced, can change and configure sharing far
faster with vi, than one can do it with the Windows control panel stuff.
As for instructing someone how to do it over the phone? With Linux, I
can just do it for them faster than I can explain, and then just tell
them to take a look at what I did.
--
Mohawk Software
Windows 9x, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
"We've got a blind date with destiny, and it looks like she ordered the
lobster"
------------------------------
From: "ismet" <[EMAIL PROTECTED]>
Subject: Re: Which Linux and which hardcopy?
Date: Sun, 23 Apr 2000 04:38:13 +0800
go redhat6.1 - i have seen all, it's easyest to install and maintain.
<[EMAIL PROTECTED]> wrote in message
news:8dg5kq$qog$[EMAIL PROTECTED]...
> I have recently purchased a second hard drive with the intention of
> installing Linux as an additional operating system. I am currently
> thinking of Mandrake, Red Hat or SUSE, and would also like to purchase
> some hard copy documentation. I have extensive experience with operating
> systems, but am a Linux virgin. I don't have the patience for books that
> spend three chapters explaining the basics; I'm looking for books that
> assume a computer background and quickly get down into the trenches. I
> will probably be doing software development on Linux, for education if
> nothing else. Given that, what advice can you give on my choices? Is
> there a chart anywhere comparing what is included in the various
> distributions? Would I be better off wait for the 2.4 kernel?
>
> I'd prefer a reply here, but you can send anything not of general
> interest to my acm dot org address (shmuel) and it will reach me once my
> computer comes back from the shop.
>
> Shmuel (Seymour J.) Metz
> ABM
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To:
comp.os.ms-windows.advocacy,comp.sys.mac.advocacy,comp.os.ms-windows.nt.advocacy,alt.flame.macintosh
Subject: Re: which OS is best?
Date: Sat, 22 Apr 2000 15:45:55 -0500
On Sat, 22 Apr 2000 16:25:04 -0400, mlw <[EMAIL PROTECTED]> wrote:
>This is the point of my argument, easier to use is NOT easier to learn.
Ah; Jedi's (and the reason I replied to the comments) said it didn't
require man pages and was easy. Both are completely false.
>A person, sufficiently practiced, can change and configure sharing far
>faster with vi, than one can do it with the Windows control panel stuff.
So use the CLI in Windows. cacls and net commands are both fully
alive and well in Win2000, among many others.
>As for instructing someone how to do it over the phone? With Linux, I
>can just do it for them faster than I can explain, and then just tell
>them to take a look at what I did.
Over the phone, with no network?
I give up. We're obviously coming from two completely different
paradigms here.
------------------------------
From: Joe Kiser <[EMAIL PROTECTED]>
Subject: Re: on installing software on linux. a worst broken system.
Date: Sat, 22 Apr 2000 16:46:40 -0400
test@myhome wrote:
>
> lets talk a little about the broken way of installing software on linux.
>
> it is most certinaly is a broken system now.
>
> a simple example. I wanted to install some rpm package
> to try some application. ok, i do
>
> rpm -Uhv foo.rpm
>
> it tells me it needs 5 others packages that are missing or not
> to the right level.
You'll love *BSD's ports system. Just cd to /usr/ports/application-name
(well, the path may differ on OpenBSD and NetBSD, but this is what it is
on FreeBSD), type make all install, and everything else is automated.
The system downloads the source code for the program, as well as
anything that the program requires to build, and finally, builds and
installs the program on your machine. Cool stuff, I haven't seen
anything comparable to this on any Linux distribution yet.
--
-Joe Kiser
Email: [EMAIL PROTECTED]
WWW: http://www.mindspring.com/~joekiser/
"Without darkness, there is no light."
-??
------------------------------
From: Mig Mig <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.ms-windows.advocacy,comp.sys.mac.advocacy,comp.os.ms-windows.nt.advocacy,alt.flame.macintosh
Subject: Re: which OS is best?
Date: Sat, 22 Apr 2000 23:01:09 +0200
[EMAIL PROTECTED] wrote:
> On Sat, 22 Apr 2000 04:47:14 GMT, [EMAIL PROTECTED] (Jim Richardson)
> wrote:
>
> >>>There is certainly no *need* to consult a man page to use these tools.
> >>
> >>I'll beg to differ. How many millions of chips would have to work at
> >>a keyboard for how many millions of years before one randomly typed
> >>"linuxconf"? WinXX's sharing is far, far easier.
> >
> >Um, just scroll through the menus till you come to linuxconf?
>
> Which is more likely to happen - right clicking on a drive or other
> object and seeing sharing, or finding LinuxConf in the menu (amongst
> dozens of other items), navigating through several other submenus to
> find the appropriate entries, and somehow getting it all working from
> there?
>
> C'mon...let's be real here.
You better get real... try to explain to the average user how right click
does and means.. but i do agree that its easyer
------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.ms-windows.advocacy,comp.sys.mac.advocacy,comp.os.ms-windows.nt.advocacy,alt.flame.macintosh
Subject: Re: which OS is best?
Date: Sat, 22 Apr 2000 16:58:58 -0400
[EMAIL PROTECTED] wrote:
>
> On Sat, 22 Apr 2000 16:25:04 -0400, mlw <[EMAIL PROTECTED]> wrote:
>
> >This is the point of my argument, easier to use is NOT easier to learn.
>
> Ah; Jedi's (and the reason I replied to the comments) said it didn't
> require man pages and was easy. Both are completely false.
>
> >A person, sufficiently practiced, can change and configure sharing far
> >faster with vi, than one can do it with the Windows control panel stuff.
>
> So use the CLI in Windows. cacls and net commands are both fully
> alive and well in Win2000, among many others.
>
> >As for instructing someone how to do it over the phone? With Linux, I
> >can just do it for them faster than I can explain, and then just tell
> >them to take a look at what I did.
>
> Over the phone, with no network?
>
> I give up. We're obviously coming from two completely different
> paradigms here.
I have used Windows, I have been a Windows developer since version 1.x.
We are not coming from "different" paradigms. My opinion, and like it or
not, we are speaking of opinions here, is that Windows is "harder" to
use than Linux. I can do stuff easier with Linux than I could ever do
with Windows, and I have ALL the windows tools. I have always had ALL
the Windows tools, right from Microsoft, and Linux is still easier.
People, for whom I've set up linux, are amazed at how much better Linux
is. Yes they miss some things, but for the most part they almost all
giggle out load about how they no longer have to reboot all the time. It
took one friend about a month or two to stop thinking "windows" and
start thinking Linux, now when has to do something with his kids
computer he is amazed at how non-sensical and backward Windows is. He is
seriously thinking about moving his kids computer to Linux because he
does not want to re-install Windows anymore.
So, lacking any reliable and rational documentation that can prove that
Windows is, in fact, easier to use, I have to say from my own
experiences, as well as watching friends adopting and getting used to
the new OS, that Windows is harder to use than Linux.
--
Mohawk Software
Windows 9x, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
"We've got a blind date with destiny, and it looks like she ordered the
lobster"
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Sell Me On Linux
Reply-To: [EMAIL PROTECTED]
Date: Sat, 22 Apr 2000 21:01:49 GMT
Centuries ago, Nostradamus foresaw a time when SeaDragon would say:
>On Sat, 22 Apr 2000 08:11:38 -0400, mlw <[EMAIL PROTECTED]> wrote:
>
>>No one in the world cares about integer arithmetic, it is a direct
>>function of the cpu clock.
>
>Incorrect. For example, the EV6 has roughly double the performance of
>EV5 in integer performance at the same clock speed. Most benchmarks
>increase linearly with clock speed. Most normal work that most people
>do is integer work.
... Which goes to display that it's not necessarily the efficiency of
the _CPU_ which is at stake, but rather that of other levels, such as
the memory bus, I/O subsystems, and such...
>>What matters is I/O and floating point.
>
>I/O, yes, floating point, huh? You think all of the corporate customers
>you run web sites, run word processors, do programming need floating
>point? What are you talking about? It's only application is in simulation
>which is in the scientific and engineering fields.
>
>>The x86 hardware has some of the worst I/O in the industry.
>
>Proof? FYI, Willamette will have a 400 MHz bus which will be the
>fastest in the industry. K7 already has the same bus frequency
>of the EV6.
Ah, but this isn't why Sun sells a whole lot of high end boxes.
Sun _doesn't_ sell expensive boxes because UltraSPARC is computationally
impressive.
They sell them because the _overall system,_ which means everything from
CPU to disk drive, is "impressive." They might be able to build boxes
using Williamette CPUs that provide similarly impressive results; nobody
has been doing so thus far.
>>Right now they have inched up to a 133 mhz front side bus. That is very
>>slow when compared to other systems.
>
>The EV6 has a 200 MHz FSB. Same as K7! So how is the K7's bus "very slow
>when compared to other systems"? I am extremely interested in hearing
>this, especially which systems you are talking about.
>
>>So, while the CPU may be clocked at 800 mhz, it can only read data from
>>ram at 133mhz.
>
>No shit sherlock? You mean all of those 1 GHz machines should be using
>1 NS memory? Where do I buy it? Man, get a clue. Have you ever heard of
>a thing known as "cache"?
Cache doesn't matter if it's getting flushed continually due to there
being disk I/O that pushes 100MB of data through. You can't afford a
100MB cache...
>>This means that unless all your processing can take place in cache,
>>your effective CPU clock is 133 mhz.
>
>But in practice all execution is done in cache. The only time bus
>speed matters is for an L2 cache miss. That's why EV6 has 4 MB
>of L2!
And if there's a "hot spot" larger than 4MB large, what then?
>>This problem affects all sorts of processing tasks for which one would
>>use a server. Ripping through a 5 meg buffer of floating point data will
>>be much slower on a PC than on an alpha. So for tasks like image
>>processing, crunching large floating point arrays, and tasks which use a
>>lot of memory, the x86 architecture is very slow.
>
>All good in theory, except for the fact that X86's do indeed have equal
>(and in the future, higher) bus bandwidth than the RISC chips (c.f. K7
>and EV6).
Which betrays the error of thinking that RISC versus CISC is really of
any relevance in the marketplace anymore... It hasn't been, for at least
five years...
>>The next thing is the PCI bus. Most x86 system are still only using 32
>>bit PCI at 33mhz. Suns have 64 bit PCI running at 66mhz. A Sun can get
>>4x more data from its PCI cards than can an x86. This means that a
>>ULTRA SPARC running at 450 MHZ using gigabit 64bit ethernet cards will
>>out perform a 800 mhz x86 using 32 bit ethernet cards because of the PCI
>>bus width and the I/O design of the system.
>
>??? What the fuck does PCI have to do with the CPU? This is all handled
>by the chipset and has absolutely nothing to do with the CPU. Are you
>arguing systems, or CPU's?
Systems, certainly. And when the data has to be transferred between
CPU, bus, and physical device, that _certainly_ has something to do with
the CPU.
>>I think you need to read up a bit about computer design.
>
>I design computers for a living, bud.
>
>>I have no idea what you are ranting about. Configuration files are read
>>at startup, and are not typically read during operation.
>
>You are also extrememly ignorant of how Linux works. Whever you do an
>ls -l, it reads /etc/passwd, patietntly parsing through this gigantic
>text file one byte at a time, looking for info. If you delete /etc/passwd,
>you when you do ls -l, you will see GIDs and UIDs instead of user names.
It does, indeed.
Which is why, for those few configuration files that tend to actually
grow to substantial size, there tends to be alternative access mechanisms.
/etc/passwd is one of the few config files where growth to large sizes
tends to occur, and thus, there is an alternative mechanism whereby the
file gets "compiled" into a DBM file so that getpwent() becomes an O(1)
operation.
The notion that this is a particularly fragile mechanism seems to me to
be confined your head. [Actually, that's not _quite_ true. There are
periodic visits by mindless WinVocates that seem to think that the fact
that the Windows Registry has a "transaction commit" mechanism means
that the fact that it layers a fragile binary data structure on top
of a filesystem makes it more robust than text-based schemes that use
the strengths of the filesystem... Those of us that actually work with
serious transaction processing systems know that simply having some notion
of transactions doesn't mean that the overall system will be robust...]
I would suggest, to the contrary, that, _particularly_ if the "compile
to DBM" mechanism is used, that this is a remarkably _robust_ mechanism,
as it clearly separates updating from reading. Many processes consult
/etc/passwd; what they "hit" is a DBM file that is _inherently_ static,
and regenerated from the master file when needed.
Nothing except the "recompile" process changes that DBM file, which is
rather different from the situation with the way the Windows Registry
joins together things that shouldn't ever change with things that change
every _second._
--
"Lumping configuration data, security data, kernel tuning parameters,
-etc. into one monstrous fragile binary data structure is really dumb."
- David F. Skoll
[EMAIL PROTECTED] - - <http://www.ntlug.org/~cbbrowne/linuxsysconf.html>
------------------------------
From: [EMAIL PROTECTED] (SeaDragon)
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: Grasping perspective... (was Re: Forget buying drestin UNIX...)
Reply-To: [EMAIL PROTECTED]
Date: Sat, 22 Apr 2000 21:00:45 GMT
On Sat, 22 Apr 2000 13:13:31 -0400, mlw <[EMAIL PROTECTED]> wrote:
>Boy are you out in left field. From a technology perspective, Windows is
>a disaster. I will argue any technical position you may take, debate the
>pros and cons on an aggregate basis. The single thing that NT does
>better than Linux, is processor affinity linking to I/O devices with
>SMP. This one feature has little effect in 99.9% of the applications
>which one would deploy using x86 hardware.
I never claimed in any post that Windows was superior to Linux. My only
point is that Linux sucks, technically. I am not interested in comparisons
to Windows but to the more serious operating systems.
>So, you made the erroneous remark that UNIX is technically inferior to
>UNIX, so I will let you frame the debate. You put out some facts, you
>have some right?
Facts why Linux sucks? How about a complete lack of asynchronous I/O
for starters. A completely lack of application-level exception handling.
A horrbibly inefficeint and slow string implementation. A window system
which requires a context switch (!) whenever you move the mouse. A
programming model which fixes resolves memory management through process
destruction. WHy don't the superior, more robust systems have these
problems?
>This is very untrue. Most Linux users have a great deal with computers.
I have met Linux users who are so new to computers, so unknowledgable
about how computers work, that they couldn't give an intelligent discussion
on e.g. what issues there are in SMP performance (aside from obvious
issues, such as locking shared data structures), could not give an
intelligent discussion on advanced data structures such as splay trees,
or fibonacci heaps, and do not even understand the benefits of virtual
memory (beyond the obvious applications of paging to disk, and memory
protetction). In my experience, most Linux users are compiler jocks
who think they are god's gift to programming because they know how to
unpack a tar.gz file and compile it into a binary (a task, of course,
which has nothing to do with programming, and is completely non-technical).
I find that most Linux advocates are completeley technically ignorant of
any serious technical issues, but instead harp on on legal and economic
issues. Most of these are obssessed with comparisons to Microsoft (and
not serious, robust operating systems), and cannot say anything intelligent
except vague, cliched generalizations about stability, security, and
scalability, and in most cases they do not understand the real technical
issues which bnenefit these areas.
>Different types of UNIX are different operating systems.
Phooey. All of the Unix operating systems do things the same. FOr
every interesting technical issue, all Unix's address the problem
in same (usually, wrong) way. See the above list and tell me which
Unix fixes the abive problems.
>This does not say that NT is not also an operating system, however
>something like Windows is debatable because the OS is really DOS and
>Windows is an extension, so it is debatable. But like a tomato can be
>called a vegetable, I guess Windows 9x can be called an OS.
Why the fuck are you so obssessed with Windows? DOes it make you feel
really good that you are the second shittiest operating system in the
world instead of the shittiest? Why don't you try comparising your OS
to something serious such as TOPS-20 or MVS for a change?
------------------------------
** 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
******************************
