In comp.os.linux.advocacy, DFS
on Mon, 16 Jan 2006 01:32:12 -0500
> Roy Schestowitz wrote:
>> __/ [Aragorn] on Sunday 15 January 2006 01:55 \__
>>>On Sunday 15 January 2006 02:51, Technomage Hawke stood up and spoke the
>>>following words to the masses in /comp.os.linux.advocacy...:/
>>>>>Roy Schestowitz wrote:
>>>>>>__/ [Roy Culley] on Friday 13 January 2006 01:16 \__
>>>>>>There seems to be a lot of repetition and overlap across points.
>>>>>>Nonetheless it's a good collection of Linux benefits.
>>>>>It also contains the lies and immaturity that are
>>>>>the hallmark of Linux "advocates".
>>>>you make so much of this that I begin to wonder where your loyalties
> I don't understand.
>>>His loyalties lie with Microsoft, as everyone on this newsgroup already
>> Having seen some pathetic habits of his, I wouldn't bother with him. I am
>> more inclined to verbally challenge Windows 'power users'
> What is it you think makes someone a Windows power
>> (very few of them exist),
> Many millions exist.
Hundreds of millions of Windows users exist. It is not
clear what percentage of them are "power users", but
it's clear that Windows is *the* dominant OS, worldwide,
on desktops, from Afganistan to Zimbabwe, presumably with
most of the users in the United States and in China.
Whether it will remain so is an unknown question, but for
now, I see this stalemate continuing for another decade or
so, with X to be eventually replaced by an X/Win32 hybrid,
incorporating the best of both worlds. (We already have
WinE, so we're halfway there.)
>> whose knowledge of Linux extends beyond a 5-minute spin with Red Hat
>> 9. Notice that the challenger here did not confute anything in the article,
>> but targetted personality instead.
> If you insist:
> (1) Linux is free of cost.
FALSE! Linux is *not* free of cost, never has been.
This may sound ridiculous but remember that TCO encompasses
multiple categories; the initial acquisition/licensing
cost is not the only cost to consider when purchasing
and/or maintaining a system. Of course the hobbyist
can download a version of Linux without too much
overhead (but even downloading and storing an image
costs money -- about $0.70 - $1.00 a gigabyte, judging
from hard drive prices I've seen recently).
The same, of course, is true in Windows, and let's
be fair here; Windows has the burden of having to be
deloused on a regular occasion, for whatever reason.
Linux has far lower antiviral costs. (They're not zero.
The CERT advisories of Linux are a cost reflected in the
user community and there's the occasional distro-specific
bug such as Li0n; some cross-fertilization bugs are also
available, spawning a market for Linux systems to prefilter
out Windows malware. How one properly accounts for these
costs, I for one don't know.)
No, Linux has not been free of cost. In any event, Linux
is not free as in libre either -- though the reasons are
a bit murkier. Briefly put, however, Linux is under
the control of Linus Torvalds and company, who will be
happy to take suggested kernel patches and modules. But
remember that they are only suggestions.
The only way to be truly free is to design one's own
utilities, although Linux isn't too bad in that regard,
since Unix philosophies and such allow for many flexible
pipelining methods and simple calls. Microsoft's
methods are a lot more complicated in that regard,
though hacks such as Cygwin are readily enough available.
> (2) Source code is available and can be modified.
> True. But so few people can or want to futz with
> the source that it's not at all a reason to switch
> to Linux.
I am forced to agree here. As a business case, the
availability of the sorce code does not in itself make
for a compelling reason. However, one can consider it,
and the good news is that, in the case of a vendor dying,
one at least has the code to give to another vendor for
modifications. It's not something, however, that can
be touted up in a return-on-investment computation.
Also, vendors don't die all that often. Microsoft in
particular is quite healthy, although a bit pudgy;
that $50B in *cash* could be used to purchase additional
companies and make it into a true multimedia conglomerate,
to increase revenue and possibly profits.
Whether the government or the users would be happy in
such a case, I for one don't know. (I would not be.)
However, a cursory scan indicates that Sony is barely
15% of Microsoft's size. Even IBM is only half as big.
Microsoft won't die unless it becomes illegal. The risks
in such a course are enormous long-term, mostly because
what the government deems illegal tomorrow might be *YOU*.
Or me. Or any of us. It had better be a darned good reason.
> (3) High quality support for Linux is free
> As it is for Windows.
However, in the Windows case it's bundled into the
product cost. Many Linux distros will give one
support if one purchases the product -- that is
what one is paying for. But support costs money.
Or, if one prefers, energy, time, and desire.
> (4) There is little possibility that support for
> Linux will be discontinued at some future date.
> This is bullshit. Suse dropped official support
> for 8.0 after only 2 years. On the other hand, MS
> is still officially supporting Win98 nearly eight
> years after its release.
> For many Linux apps, death of the developer would
> mean death of the app. There's little to no
> guarantee support for a given OSS app will be
> around tomorrow, let alone a year from now.
This is where the cost of support comes in.
An application is now dead. ("Ming" is one such
application, BTW.) Who picks it up?
The same could be said for a number of Windows
applications, of course.
> (5) No major obsolescence, planned or otherwise,
> with Linux.
> This is laughable, really. Not only does Linux
> have planned obsolescense in those distros on a
> fixed-release cycle, the amateurish nature of
> Linux/OSS development means apps are constantly
> being updated and previous versions are dropped
> from support.
I'm not so sure of this one. The entire software
management problem is giving me headaches (yes,
Gentoo's being a pain again, but part of it is
my fault; ACCEPT_KEYWORDS=~x86 takes one out of
the stable realm). Whenever package A is upgraded,
any package B depending on it may be affected.
> (6) There are no forced upgrades for Linux users.
> Nor are there any for Windows.
False on both counts. Or has anyone noticed security
upgrades for the 2.0 kernel series lately? How
about kernel 1.2.13 or 1.0.9? Of course, nobody's
forced to upgrade -- but then nobody's forced to
own a computer, either; it's just more efficient
nowadays to do so.
> (7) Linux upgrades are free.
> But there's no guarantee that an upgrade will be
> forthcoming. So it's meaningless.
And it's false anyway; upgrades cost bandwidth and
storage space. Of course most people won't think of it
in that fashion.
> (8) Don't have to keep track of Linux licensing.
> This is a silly non-issue. It's no more difficult
> to keep track of Windows licenses than it is Linux
Gentoo keeps careful track of licenses, when it has to.
A fair number of licenses are in /usr/portage/licenses;
there are 745 entries on my system. Of course one
of them is /usr/portage/licenses/GPL-2; another is
Whether *I* can keep track of all these, I'm not sure.
> (9) Linux features superior security.
Security can only be guaranteed by the sysadmin, and
good practices thereby. Linux by default might have
better security out of the box but one cannot guarantee
that a virus won't wander in a door the sysadmin has
The good news: Linux features superior security *by default*.
But any good admin can harden a Windows installation -- if
only by installing a front-protecting firewall blocking
most of the malware probes.
> (10) Linux is highly resistant to system crashes
> and rarely needs rebooting.
> I haven't used it enough to confirm this, but I do
> know Linux apps crash very frequently. Even in my
> limited use of Linux, I see sigseg faults very
> often. I had Kubuntu 5.0.4 running on my P3-800,
> 512 system, and Konqueror sigseg faulted at least
> every half hour when I was using it heavily.
> A program called Quantas would raise a sigseg
> fault each time I tried to close it.
> My WinServer2003 system is very solid. It has
> only ever crashed due to an issue with Half Life 2
> and the onboard video.
There are crashes, and there are crashes.
 Kernel panics/Blue Screen Of Death. On both
systems, these are rare. Windows has a minor
advantage here in that such failures are distressingly
 Library failures. These might be construed
as a call to a library doing something unexpected.
For example, most people can't fopen("/etc/passwd","a").
 Program faults. In most cases this is because of a bad
pointer. Depending on system the program may just die
horribly, put up a requester that says "I'm dead!", or
try to handle the exception and keep on going, in some
cases sans appendages required to keep it going.
Windows has a slight advantage here, if only because
a dying program's "death knell" can be intercepted
by the OS.
 Hangs. The program is either waiting for something else
to do something but that something else hasn't been told
it needs to, or has entered into an infinite loop which
has no exit.
There's no real difference between Linux and Windows here,
as far as I can tell from the documentation. From my own
personal experience IE is unpredictable, but then Microsoft
is encouraging users to switch to Firefox anyway.
> (11) ...users often find that all the applications
> that they want are freely available on the
> Internet and that it is no longer necessary to
> purchase any commercial software.
> Probably true for some home users. Users with
> high multimedia requirements and gaming are going
> to be left out in the cold with Linux. And every
> large business has desktop computing requirements
> for accounting and business data analysis that
> Linux apps cannot satisfy.
I'll admit to wondering about this. _Spaceballs_ in particular
threw me for a loop, but I fought back by re-emerging.
(It had to do with AC3 big-endian encoding. Mplayer in the
stable version apparently didn't support that -- though it's
possible it does now. It certainly does now on my system.)
Rosegarden has been a thorn on my side, until I found a
workaround. Apparently there's an issue with supporting
artsd and alsa simultaneously or something, but that's now
resolved as well.
> (12) Linux has hundreds of distros.
> True. Linux nuts call it choice and diversity.
There is only *ONE* Linux. ONE. Let me repeat that to be
perfectly clear and that there are no misundersatndings:
there is only *ONE* (1) (un) (uno) (eins) Linux.
The rest is window dressing.
I should note here that it is possible to build
the "window dressing" around other kernels. An
interesting possibility, for example, is building
KDE on Windows; Gentoo should be buildable around
the HURD kernel, and Debian already has a release
out around that kernel. FreeBSD has some quirks
but KDE and Gnome are already thereon. I don't
see why one couldn't port KDE and Gnome to any
sufficiently capable UNIX(tm) platform, either.
And yes, that includes Windows -- which was certified
under the UNIX98 standard some time ago, though one
might have to acquire additional software (IIRC, the
MKS package, and perhaps SFU as well).
> (13) Linux features a high degree of flexibility
> and customization.
> True. This is a very attractive feature of
False. This is not an attractive feature of Linux/OSS.
Unless one likes flexibility (I do :-) ). See your answer
to (12), above. You're being inconsistent.
> Also, it's not difficult to configure
> WinServer2003 as a file server, print server, app
> server, mail server, terminal server, VPN server,
> domain controller, DNS server, DHCP server,
> streaming media server, etc.
This speaks well of Win2003's flexibility; however,
most people will be using Windows XP Home Edition,
which does not offer most of these features.
But then, how many people at home need a DHCP server?
(I only need one because I at one point wanted to
get my ancient Sparc to do something interesting.
Admittedly, I've now completely botched it, though
I think the hardware still works.)
> (14) Linux and other free software uses open file
> The world has gotten along fine with closed
> formats for 35 years, which shows it's a non-issue.
The world has gotten along fine without computers for
far longer than that.
> (15) Linux is generally faster for a given set of
> hardware specifications. This is due to greater
> optimization of the source code, including far
> less code bloat.
> While Linux code sizes are smaller than Windows,
> the apps do NOT run faster on the same hardware.
> They're usually significantly slower.
Benchmarks are hard to come by. My general feeling,
at least on $EMPLOYER's hardware, is that Linux sails
along fairly smoothly but Windows XP Professional Edition
wallows like a stuck pig. This may be in part because
of usage models, as I simply leave Linux up here,
whereas I have to shut down and reboot this system
to run Windows.
But it's the same hardware.
> (16) Linux features a high degree of compatibility
> with other operating systems.
There are some issues regarding compatibility and
it depends on the metric. I'm heartened, however;
I can run IE 6.0 without trouble on my Linux/WinE
system; the instructions for setup are simple.
However, I have so far completely failed to set up
Visual Source Safe on this same system. Of course
part of that is demand ("uh, what's VSS, again?");
VSS is something used by software developers, whereas
IE is used by just about everybody.
> (17) Linux/OSS vendors and developers have very
> high ethical standards.
False. No different than any other group of humans.
The main difference: many Linux programmers work out
of love -- they are amateurs, in the truest sense.
However, many of them are also paid computer professionals
with years of experience.
> As long as 'ethical standards' don't include the
> shameless copying (blatant ripoff really) of
> features and designs of MS and other closed-source
> products, including other operating systems (like
> BeOS), I would tend to agree.
Still no difference; Windows has shamelessly ripped
off Xerox PARC in the past, for example, and is
probably real interested in copying over the
Apple GUI portions it finds useful.
This is par for the course, and a truly innovative
interface would find few adherents anyway. Everyone
knows what a "window" and a "menu" is, though.
(Well, almost everyone; some might have to be taught.
But that's what schools and parents are for... :-) )
> (18) Linux has lower hardware requirements.
> Depends on the distro.
And on the Windows environment. There's probably
an XP Embedded that might only 32 megabytes. Of
course there used to be a DOS that took at
most 64 *kilobytes*.
I'll have to see if I can install a Linux on my
4 MB 386 motherboard -- but modern Linux distros
can take gigabytes of hard drive space, mostly
because they have gigabytes of packages to install
on that space.
OpenOffice 2 in particular takes 269 megabytes on my system.
> (19) Linux runs on lots of hardware platforms.
> Another non-issue. People that are going to
> convert from Windows are going to be running x86
> systems. And before you start wailing about
> Sparcs and IBM zSeries, the article is about
> converting from Windows.
And emulators for the x86 platform are also
readily available. Windows can run *everywhere*,
admittedly with a performance penalty. But
VmWare is very portable.
> (20) Linux is a superior choice for use in
> academic institutions
> For teaching OS design and C programming, anyway.
False. Linux and Windows are roughly equal in this
context. The main advantage here is in initial
acquisition costs; also, malware has yet to intrude
on Linux to the degree it is perambulating around
on Windows system. But Linux isn't immune to malware,
See (9), above.
> (21) For governmental agencies, Linux and other
> free software allows for transparency of data
> because it stores the data in formats consistent
> with industry-wide standards.
> Same as #14
And not required by governmental agencies anyway.
The law might have open format requirements, but
that's because someone is attempting micromanagement
of the problem.
In any event, the Word format is very open. OpenOffice
can read it, no problem. (That the Word format is
not an established standard doesn't really matter
in this context; as long as someone can properly read it,
it's an open book. It might mean fewer headaches down
the road if it's open to begin with, of course -- but
for now, Windows has the problem of its own success to
deal with, and unless one gets into DRM/licensing/codec
space, there's not a lot that Microsoft can do to break
Word, and there's a *lot* of screaming by customers if
> (22) With Linux and other free software there is
> little reason to fear the existence of backdoors,
> in large part because all of the source code is
> available for inspection.
False. Rootkits are common (as are checkers).
In olden times, if a wall had handholds, it was rendered
less effective. Even without handholds, the enemy can
get over the wall with scaling ladders -- hence the pots
of hot oil kept on the walls to pour down on the enemy.
In modern times, however, the enemy not only has scaling
ladders, but everything's done in a bit of a fog. He
may get in the inner bailey while the King is sleeping.
> I wrote a cola post in 2004, entitled
> GNU\Linux\OSS might as well be closed source...
> "... for all you cola bozos know about it.
> Linux morons always brag about having the source
> code to your kernel and apps - as if you've ever
> examined it and know it's 100% safe. Sure.
> How do you know the binaries included with your
> distro are built with the source files that come
> with it? You don't.
I have some assurance. The main issue with Gentoo are
the compilers, and those are built frmo source too.
But the initial bootstrap has issues -- and there's not
a lot one can do about it unless one can physically splot
magnetic bits on a diskette with a micropen or something,
to get started.
There have also been attempts to hide checks in the
source code, in older compilers. These checks
spit out anomalous code when encountering specific
token sequences -- such as those one might encounter
in source code for /bin/login.
Open source is no guarantee -- though it helps. But it's
all a matter of trust.
> How do you know the code doesn't have back doors
> and trojans in it? You don't.
> You just blindly trust your OSS vendor. Now I
> know it's not OK to trust Microsoft, but blindly
> installing anything from Mandrake or Suse or Red
> Hat is fine. This is typical Linux hypocrisy.
Microsoft is one of the most trustworthy companies
out there, surprisingly enough, according to
at least one survey. This may be forced on them,
of course; if they don't do it the right way they
may very well lose their dominant position. In a
way they are -- again -- victims of their own success.
> Even if you're a C\C++ master and Gentoo user, and
> compile every app. from source and specifically
> for your machine, you haven't read the hundreds of
> thousands of lines of kernel code and code for
> every app on your machine.
> So, for all you nutcases know about the
> GNU\Linux\OSS software running on your systems
> right now, it might as well be closed source code,
> and may be vulnerable to an "NSA back door" right now.
And of course Windows is far better in this regard, as
their millions of lines of source code are audited
This argument is problematic on multiple fronts.
> (23) Using and advocating Linux helps foster a
> healthy diversity and increased competition
> throughout the software industry.
False. No more than any other system. Computers
can be used for good, or for evil; ditto for
software. Linux can be used to help foster a
healthy diversity, or enforce a highly restrictive
and secure system, letting only those who are
specifically authorized by the "old boy network" in,
and only on a "need to know" basis.
> True, and this is, to my mind, the most important
> reason to use and support Linux and other MS
> alternatives. But using and advocating isn't
> enough; you need to be spending and purchasing and
> donating. The economic world generally doesn't
> share OSS ideals; it's all about money. And
> marketing Linux costs money.
Bear in mind that MS is an alternative in this space,
and a very open system, if unintentionally so. The
main problem is documentation for such goodies
as COMMDLG.DLL, which can easily be replaced (and
in fact already has been, if only by Microsoft).
> (24) Linux and other free software have not only
> caught up with, or some cases surpassed, their
> proprietary counterparts, but they are also
> developing at a faster pace.
> I think Apple is eating MS' and Linux' lunch in
> terms of creativity and innovation.
I think you're entirely wrong; if anything, Linux is
eating *Apple's* lunch. However, Apple is doing
well enough, although I'm not entirely sure where
their revenue stream is nowadays.
In any event, Microsoft won't give up easily but they
are beset by competitors.
> (25) Linux and other free software provide the
> opportunity for users to contribute to the advance
> of software technology because the source code is
> freely available to study, improve, extend and
> Same as #2
And also a non-issue for most business users.
> (26) Linux systems don't need disk defragmenting.
> Another non-issue.
The entire question of defragmentation is an interesting one.
 try to keep the file blocks contiguous?
 try to keep the file descriptors (inodes and directory
entries in the case of Linux; directory entries in
the case of Windows) contiguous?
 try to keep the files as close to their parent directory
as they can?
 try to analyze usage patterns for files and directories,
and place things accordingly?
 say "shove it" and simply cache things where possible?
I'm forced to agree, somewhat reluctantly. However, it's clear
that part of the problem is caused by usage patterns; briefly,
if one opens a "log file" and also opens and writes to several
files, while logging, one is looking at heavy disk fragmentation
It's still legal to go .sigless.