Subject: Re: seti runs better on win or linux
From: AthlonRob
Date: 12/10/2003, 00:15
Newsgroups: alt.sci.seti

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
NotDashEscaped: You need GnuPG to verify this message

On Sat, 11 Oct 2003 13:52:38 -0700, Rich <someone@someplace.com> wrote:
My my, you need to do some research.

You need to trim your posts a bit... they're getting a bit long.  Cut 
out some of the quoted material, maybe limit it to just what you're 
actually responding to, please.  :-)

http://lists.debian.org/debian-glibc/2002/debian-glibc-200210/msg00216.html

Bug#155939: marked as done (libc6: libc6 broke apache and all its modules)

I would have picked one a bit more recent than that... maybe the GCC 
3.2-compiled Mozilla not working with Java?

Moving from libc5 to glibc was like moving from Windows 95 to Windows 
2000.  How many things broke there?

And... that bug you list above... certainly did *not* have anything to 
do with Debian-stable at the time.  It had to do with the development - 
for testers only - version of Debian.

Linux patches indeed to break things, and often fail to work.

The number of linux patches is astounding to see. There are thousands if
not hundreds of thousands of patches out there, just for linux.

The kernel, or the applications?

I think you misunderstand what a patch is... all a patch is is a change 
in a program.  In the Windows land, that usually means replacing some 
binary files.  In Linux land, it virtually always means changing some 
code.  Every time a program is changed, a patch is generated.  So of 
course there are hundreds of thousands... if not millions... of patches 
for Linux and its applications.

There are patches to add features, there are patches to fix bugs, there 
are patches to secure things.  If we just look at the kernel, there are 
patches between different versions (if I have the 2.4.21 source, I can 
download a ~7MB patch to go up to 2.4.22 instead of a ~30MB tarball), 
there are the patches for beta kernels, there are nightly patches to 
beta kernels, then there are a few (tens, I don't think hundreds, but I 
could be mistaken) 3rd-party patches for it to add features here or 
there.  I, personally, used low-latency patches whenever I compiled 
2.4.x kernels.  I also used ACPI and CPU Frequency patches sometimes to 
add functionality with my (non-APM) laptop.

Generally patches that break things are patches developers are working 
on and share only with other developers.  For an example, I was working 
with some cpufreq patches a while back that were only for developers, 
and sure enough, they did break things.  But I wanted to live closer to 
the edge than I should have been.

I wonder if instead you mean patches, you mean updates?

I don't hear of updates breaking things very often, either.  An update 
would be a binary package distributed out to replace something that's 
broken.  For instance, I updated OpenSSL the other day on my server.

These are always tested before being released, assuming you're dealing 
with a reputable distribution.  They don't break things very often at 
all.

They can only affect the  application not the OS or its operation.

You need to look into the kernel patches, they affect only the OS, which
of course, affects all the applications and users.

Kernel patches aren't something Joe Lunchbox needs to worry about, 
though.  Many people use Linux every day without recompiling the kernel 
- and if you patch the kernel, you do need to recompile it.

People update their kernels often, though, I suppose.  They don't really 
need to very often, but they tend to more often.

But how often does that break things?

The OS and applications are completely separate in linux while in MS they 
are deliberately intermingled.

Bullshit, the dll's in windows derive from shared libraries in unix.
In unix library compatibility issues are as ubiquitous as dll problems,
and for exactly the same reason. I don't know how many times I've trussed
an ailing program, notices the wrong library path being used, and fixed
it via rearranging the LIBRARY_PATH variable of the shell. But guess what,
this can break other programs. It's somewhat of a black art.

Linux (and Unix) tends to take a more layered approach to things.  I 
think the last guy was talking about this.  The GUI is independent from 
the kernel, slrn crashing won't have any effect on the kernel itself.  
In Windows, everything gets jumbled together, the GUI part of the 
underlying OS (and for servers -a GUI?  Blech!).

What the heck are you talking about with $LIBRARY_PATH?

rob@dell rob $ echo $LIBRARY_PATH
 
rob@dell rob $

I've never had to mess with that.  The only time it might come in handy 
is if you're running two seperate versions of a single library, 
compiling applications against one and then the other - but that's plain 
stupid, really.  Why would you do that?  It sounds to me like a black 
art nobody ever practices... because there is no need for it.

If you found you needed different $LIBRARY_PATH's for different 
programs, why don't you just write a quick bash script for the offending 
applications to set the variable and call the program?

    It is certainly an insane (or your favorite adjective) to mix 
applications with the OS. Just on best design practice the choice is linux.

I'm talking about linux, not an application. Not that you can run linux
applications on windows, solaris, or any other OS.

???

We're just talking about the kernel, then?  Linux is the kernel, not the 
applications that utilize it.

And you can run *plenty* of applications written for Linux on Windows, 
Solaris, or any other OS.  I just compiled a whole suite of GNU 
utilities on my Solaris shell account.  For Windows - take a gander at 
cygwin.

     http://www.ale.org/archive/ale/ale-1999-11/msg00310.html

     Linux, the official OS of Red China

You really like that article, I gather?  I wonder why you keep posting 
it, though.

Your inability to find documentation is astounding. I'm amazed that you
found any documentation with linux at all, as they don't even give you
that piece of paper that had you so confused.

How is it that you can look up information on linux but never even tried
with windows? Is this supposed to show how biased you are? It's certainly
not an advert for your intelligence.

There is much information on the net WRT Microsoft, and there are many
books available on the subject. You may be able to get them from the
library. The A+ books seem to work for me.

There is more information on the net on Linux than Windows, 
proportionally.  Linux *is* on the internet, in source and binary 
format.  Virtually all its applications are there, too, in both formats.  
Windows, on the other hand, has a few hints and tricks available.

And for Linux documentation, have a look at /usr/doc.  Mine is filled 
with 107MB of documentation, including how-to's, application 
information, and anything else you can think up.  Oh, and that's 
compressed.  Tell me you have 107MB of compressed text on Windows 
available in Help.  ;-)

Look, your claims about linux security are way out of line. Along
with your claims about linux patches. There are more patches for
linux than the source code for windows, there are patches for everything.
The kernel patches are the most troubling. Were linux as solid as
you say they would be few and far between instead of being as ubiquitous
as linux itself.

Again, you misunderstand patches.

People require different levels of security depending on the task 
they're trying to perform.  For the NSA, they need more security than 
the Linux kernel is able to provide due to export restrictions.  So they 
patch it.

Crypto stuff is still restricted when it comes to exporting it - and 
Linux is all about being non-restricted.

Windows is neither good or bad, Linux is neither good or bad (and it's
nowhere near as perfect as you claim).

Nothing is perfect, but things are definitely good or bad...  :-)

Installing linux on a PC has always been a difficult procedure, although
it's getting better. But the problem for almost everything has been lack

If you want an easy install, go try Mandrake.  Or Lycoris.  Or Lindows.  
Or Xandros.  Or LibraNet.

They're all very very simple to install.  Easier to install than MS 
Office.  As easy to install as AOL.

of drivers (and this is still true for much hardware you might buy). Many
cheap computers come with winmodems, which use a windows service to
replace hardware on the modem (and reduce the cost to $10-$20). Just try
and install one on a linux box. Your hardware choices are still more
limited (although much better than just 5 years back), and manufacturer
support for hardware is today just so-so, rather than virtually nonexistent.

Hardware support is amazing these days.  I'll grant there are scanners, 
printers, and winmodems that don't work, but even then - the vast 
majority of those *do* work in Linux, and actually work better than with 
Windows.

Manufacturer support for any drivers basically sucks - in Windows or in 
Linux.  It's always a PITA to get any real support from most all 
hardware makers.

All that matters is wether or not it works - and most of the time, it 
works.

You still need to be careful what hardware you use, just for starters.

The same is true in Windows, however.  Shoddy hardware = an unstable 
system.  That's less of a case in Linux, however.  Sure, you need to 
make sure your hardware works... but it does most of the time.  I 
haven't checked before buying most of my hardware, honestly, and haven't 
had any issues.  I have a fair variety of hardware, too.  :-)

-- Rob | If not safe, Email and Jabber: | one can never be free. athlonrob at axpr dot net | -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQE/iI87hm6KEoOOAe0RAo2nAJ4uiz0i3hUWQaPKQaDBWZ8pz6vvIgCeOUcz s+B7MqTM/NgjzENP2Kueb40= =rdq4 -----END PGP SIGNATURE-----