Linux-Advocacy Digest #481, Volume #31 Mon, 15 Jan 01 10:13:04 EST
Contents:
Re: Benchmark tests - who cares? (Ian Davey)
Re: Linux 2.4 Major Advance ([EMAIL PROTECTED])
Re: OS-X GUI on Linux? (.)
Re: Why Linux won't get far in Luxembourg's comapanies. (A transfinite number of
monkeys)
Re: Red hat becoming illegal? (.)
Re: Benchmark tests - who cares? (sfcybear)
Re: Linux is crude and inconsistant. ([EMAIL PROTECTED])
Re: One case where Linux has the edge ([EMAIL PROTECTED])
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Ian Davey)
Subject: Re: Benchmark tests - who cares?
Date: Mon, 15 Jan 2001 14:18:46 GMT
In article <[EMAIL PROTECTED]>, Donn Miller <[EMAIL PROTECTED]>
wrote:
>
>Basically, I'm wondering if some people in here are "bandwagon" Linux fans.
>You know, they like it just because it performs well on benchmarks. If
>FreeBSD scored 100 worse than Windows on a webbench result, I would keep on
>lovin' my OS! It's because I know that it is an open source product, my
>OS would have a tremendous future ahead of it.
I think it's more of a reaction to all the benchmarks that have been used
against Linux here in the past by the obsessive Windows users who keep
posting in COLA. It's the old equivalent of rubbing their faces in it. Not
that it's really necessary of course, but at the very least it might shut some
of them up for a while.
ian.
\ /
(@_@) http://www.eclipse.co.uk/sweetdespise/ (dark literature)
/(&)\ http://www.eclipse.co.uk/sweetdespise/libertycaptions/ (art)
| |
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: Linux 2.4 Major Advance
Date: Mon, 15 Jan 2001 14:06:20 GMT
Jan,
> I just don't know that certain. But what I do know is that, yes, IIS was
> generating the content.
to continue the detective work, Microsoft's SWC/TWC source code contradicts
this claim. Check out DEFINE.H in Microsoft's TWC/SWC Spec source code
package, it says:
#define SERVER_SOFTWARE_NAME "TWC/SWC 3.0"
in DYNAMIC.C you can see how the TWC/SWC code creates a HTTP reply [?] in the
InitSpec( ) function:
QueryServerSoftware(ServerSoftware);
So, according to Microsoft itself, the web server used in this test was
"TWC/SWC 3.0". It was probably changed to "IIS 5.0 with SWC 3.0, API: TWC
3.0" later on, for marketing reasons. This was done perhaps to avoid uneasy
questions: "why is IIS 5.0 so slow? Is Microsoft going to fix all performance
problems by moving benchmarks into the kernel?". Exactly those questions are
being asked here and now :-)
Thomas
> > the Windows 2000 kernel in kernel-mode, and that almost no (if any) IIS
> 5.0
> > code was executing during the benchmark. This was my point from the
> beginning
> > :-) Your claim is that SWC is caching full static HTTP replies. Then it's
> in
> > blatant violation of HTTP RFC's, IIRC the 'Date:' field for example has to
> be
> > generated for every request. I just digged RFC-2616 up, it says: "Origin
> > servers MUST include a Date header field in all responses, except in these
> > cases: [... hardware has no clock ...].") So SWC 3.0 either violates the
> HTTP
> > RFC's (and thus Spec rules) in a spectacular way, or it generates HTTP
> fields
> > autonomously (and thus qualifies as a webserver). Pick your favorite.
> >
> > Thomas
> >
> > > content. It serves up cached copies of static pages. That's what it
> does.
> > > Period. It is impossible for SWC to produce a dynamically generated
> page.
> > > Period. The sooner you get over these facts the sooner you can rejoin
> > > reality.... sheesh...
> > >
> > > Then again - thinking about it... ok, so what? Say SWC is some
> mysterious
> > > here-to-unknown product MS has that no one has noticed until it went
> > > head-to-head with the linux kernel mode webserver and THEN, desperate
> for
> > > answers why linux only was a scant 2.7% faster the zealots had to go
> digging
> > > for some exuse. Amazing that no one else has noticed this interesting
> > > product that can do such miraculous performance and is tucked into the
> > > kernel yet multi-million dollar players have simply "missed" it -
> whoops,
> > > just like that. But mcnash spots it by his own mind-reading
> interpretation
> > > of the source code to a benchmark.
> > >
> > > I do see that by examining the files dell submitted for the tux results
> that
> > > there is a line that reads: "interact with the TUX kernel subsystem" -
> there
> > > we have it, proof that tux is running in the kernel space. There is
> > > documentation for how to access it from user space too. So, there you
> have
> > > it... tux in the kernel... whatever...
> > >
> > > silly
> > >
> > > <[EMAIL PROTECTED]> wrote in message
> > > news:93q1j7$nhu$[EMAIL PROTECTED]...
> > > > Contrary to your assertion, SWC is in the kernel, it's visible from
> > > > user-space as a Windows 2000 device. Proof is Microsoft's own
> submitted
> > > > source code:
> > > > http://www.spec.org/osg/web99/results/api-src/Dell-20001212-TWC.zip -
> > > unpack
> > > > it and open the twc.c C-sourcecode file. Search for 'SWC', it gives
> this
> > > > comment: "// Open Kernel SWC device": Q.E.D. Moreover, try searching
> for
> > > > 'IIS' or 'ISAPI' in the whole source-code package - you will find only
> > > one!
> > > > You will find many references to 'SWC' (Microsoft's in-kernel
> webserver)
> > > and
> > > > 'TWC', the API to this in-kernel webserver. You will even find some
> > > interface
> > > > definitions in twc.h. If you ever programmed dynamic applications
> (ISAPIs)
> > > > under IIS, you'll immediately recognize that in this benchmark no IIS
> was
> > > > used for the dynamic requests. (maybe IIS was used for the 0.005%
> CGI's
> > > > SPECweb99 generates.) Calling the test-results 'IIS 5.0 + SWC 3.0' is
> most
> > > > likely a boldfaced lie, or at best an extreme exaggeration. In reality
> it
> > > was
> > > > a "99.99% SWC 3.0 + 0.005% IIS 5.0" test. Thomas
> > > >
> > > > > <[EMAIL PROTECTED]> wrote in message
> > > > > news:93mbpa$p17$[EMAIL PROTECTED]...
> > > > > > Jan, if it quacks like a duck and walks like a duck, then it's
> very
> > > likely
> > > > > a
> > > > > > duck. Microsoft's own in-kernel SWC 2.0 web page (the outdated SWC
> > > > > version)
> > > > > > at http://www.microsoft.com/TechNet/iis/swc2.asp says that this
> > > 'front-end
> > > > > > cache' accepts and answers web requests, logs those requests into
> its
> > > own
> > > > > > separate binary logfile, and supports only the HTTP 1.0 protocol.
> The
> > > > > > Microsoft SWC 3.0 SpecWeb99 submission webpage (I couldnt find
> > > information
> > > > > > about SWC 3.0 anywhere else) at
> > > > > >
> > >
> http://www.spec.org/osg/web99/results/res2000q4/web99-20001211-00082.html
> > > > > > says that SWC 3.0 has its own dynamic API as well: "TWC 3.0". If
> this
> > > > > > in-kernel web-thing accepts web requests, serves web requests,
> logs
> > > web
> > > > > > requests and provides ways to write dynamic webpages, then it's
> what?
> > > A
> > > > > > webserver. Surprisingly, the SpecWeb99 benchmark (check out the
> > > functional
> > > > > > specification at http://www.spec.org/osg/web99/) needs these
> webserver
> > > > > > features, little more. I repeat, from the submission page it's
> pretty
> > > > > clear
> > > > > > that little if any IIS 5.0 code was running in this test - nothing
> > > makes
> > > > > this
> > > > > > more apparent than the fact that no IIS 5.0 tuning was done at all
> on
> > > this
> > > > > > system! For example compare it with the IIS 5.0 tunings done in
> the
> > > > > > following, much much slower 4-CPU SPECweb99 Windows 2000 / IIS 5.0
> > > result:
> > > > > >
> > >
> http://www.spec.org/osg/web99/results/res2000q2/web99-20000612-00049.html
> > > > > .
> > > > > > This submission page is full of IIS 5.0-specific tunings, while
> the
> > > SWC
> > > > > 3.0
> > > > > > submission has none at all! IIS 5.0 was probably just taking away
> some
> > > > > space
> > > > > > on disk and RAM, and was idling around - this was probably the
> best it
> > > > > could
> > > > > > have done to help get a better result ;-) Obviously this is not
> what
> > > > > > Microsoft PR wants us to believe though :-)
> > > > > >
> > > > >
> > > > > You are really dense aren't you? SWC is a web CACHE - do you know
> what
> > > the
> > > > > word cache means? Do you understand how a web cache works? Obviously
> > > not.
> > > > > Where do you think the pages the cache is supplying were
> generated?????
> > > Do
> > > > > you think the cache created the pages??? HELLO???!!! Doh!!! IIS5
> created
> > > the
> > > > > pages and if a static (keyword) page was requested again and it
> hadn't
> > > > > expired it was served by the cache and not by IIS, all the dynamic
> pages
> > > > > were served by IIS5 time and again.
> > > > >
> > > > > I mean, really - you enter a technical conversation without any
> > > > > understanding of how a web server and/or cache works and expect us
> to
> > > read
> > > > > that crap? Gee - did you think that no one at specweb would notice
> > > something
> > > > > clever like, gee, they didn't use a web server, they just served up
> > > > > pregenerated and cache pages (amazingly they have time travel worked
> out
> > > so
> > > > > they could pregenerate even the dynamic content to serve up from the
> > > cache).
> > > > >
> > > > > And, SWC does not run in the kernel, neither does IIS5.
> > > > >
> > > > > (not that I care really, I only make fun of Tux being in the kernel
> to
> > > > > remind linux loosers about how much they tried to give NT advocates
> > > grief
> > > > > because NT runs speed critical components in the kernel - nice to
> note
> > > that
> > > > > tux/linux is merely acknowledging the NT method of doing things as
> the
> > > best
> > > > > and copying it, like they have copied everything else).
> > > > >
> > > > >
> > > >
> > > >
> > > > Sent via Deja.com
> > > > http://www.deja.com/
> > >
> > >
> >
> >
> > Sent via Deja.com
> > http://www.deja.com/
>
>
Sent via Deja.com
http://www.deja.com/
------------------------------
From: [EMAIL PROTECTED] (.)
Subject: Re: OS-X GUI on Linux?
Date: 15 Jan 2001 14:28:14 GMT
Edward Rosten <[EMAIL PROTECTED]> wrote:
>>> It would be easier making X run on top of Quartz than on top of nothing,
>>> since X would not need drivers (it uses the drivers through Quartz),
>>> much like, say eXceed on NT.
>>
>>
>> X already has drivers for Apple hardware coming out of its ears. It works
>> just fine. The only sticking point would be the port, which to my knowledge
>> has already happened.
> But if you're running X on top of Quartz, you would go through the
> quartz API, and not access the hardware directly.
Theoretically you could shut off quartz and run X solo.
>> Though I cant think of one good reason to run X on top of quartz.
> I can think of plenty.
> It would mean that you could port other UNIX GUI programs easily
Honestly, I cannot think of one UNIX GUI application that doesnt have at least
an equivalent under OSX.
> it would mean that you could put things on your display from other machines.
Which nearly everyone does not need. If you want that sort of "server"
functionality, just use real unix.
> Personally I'd rather Quartz ran on top of X. That way, all Quartz apps
> would have remote display capabilities, which for me would be preferable.
Ever look at an mpeg movie streamed through X to a server installation?
Its terrible.
=====.
------------------------------
From: [EMAIL PROTECTED] (A transfinite number of monkeys)
Subject: Re: Why Linux won't get far in Luxembourg's comapanies.
Date: Mon, 15 Jan 2001 14:34:44 GMT
On Sun, 14 Jan 2001 21:33:50 +0100, Bartek Kostrzewa <[EMAIL PROTECTED]> wrote:
: My friend's father has a small company, he asked me to give him a proposal
: for a file server (serving 8 computers with 500MB/day/PC), so I built a
: server for 1500$ with SCSI, AMD Duron 750, 256 MB of RAM and a 100 Mbit NIC,
: of course, I told him I'd install Linux and set up SAMBA for file serving
: (the company is 100% M$ based). When he heard the price he said: "What?
: That's far too cheap! I need to spend at least 7500$ on it, so I can reduce
: my tax charges at the end of the year!" Now he bought a Win2k Server based
: Compaq Proline server powered by an 933Mhz PIII, 256MB of RDRAM and 60GB
: RAID-10 (4 30GB 10K rpm SCSI harddrives in RAID mode, stripped and imaged
: together).... and that for 8 computer low-profile file-sharing.
Ok, he needed to spend at least a certain amount, and you couldn't manage
to do that? You need a serious upgrade to your spec'ing skills.
: Even with the maximum service option possible (RedHat) the Linux-based
: soultion wouldn't have cost enough...
All well & fine, but look at the *hardware* cost differences between your
Duron box, vs. this Compaq machine. Heck, I know of *TAPE DRIVES* that can
add $5000 to the bill. Don't blame your own failure on Linux.
--
Jason Costomiris <>< | Technologist, geek, human.
jcostom {at} jasons {dot} org | http://www.jasons.org/
Quidquid latine dictum sit, altum viditur.
------------------------------
From: [EMAIL PROTECTED] (.)
Crossposted-To:
alt.destroy.microsoft,comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Red hat becoming illegal?
Date: 15 Jan 2001 14:41:06 GMT
In comp.os.linux.advocacy Chad Myers <[EMAIL PROTECTED]> wrote:
> "The Ghost In The Machine" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]...
>> In comp.os.linux.advocacy, [EMAIL PROTECTED]
>> <[EMAIL PROTECTED]>
>> wrote
>> on 14 Jan 2001 20:39:41 GMT
>> <93t2qd$gob$[EMAIL PROTECTED]>:
>> >In comp.os.linux.advocacy Chad Myers <[EMAIL PROTECTED]> wrote:
>> >
>> >> "Chad Myers" <[EMAIL PROTECTED]> wrote in message
>> >> news:Rrj86.2343$[EMAIL PROTECTED]...
>> >>>
>> >>> "Giuliano Colla" <[EMAIL PROTECTED]> wrote in message
>> >>> news:[EMAIL PROTECTED]...
>> >>> > Chad Myers wrote:
>> >>> > >
>> >>> > [snip]
>> >>> > >
>> >>> > > Hmm, oh well. Never had a reason to really. The past two jobs I've
>> >>> > > worked at, Linux couldn't be used AT ALL because of all it's
>> >>> > > shortcomings, so this "option to be configured" really doesn't
>> >>> > > mean dittly squat.
>> >>> > >
>> >>> >
>> >>> > Where did you work? At a gas pump?
>> >>>
>> >>> 1.) Video people did tons of video editing with files well over 2GB.
>> >>> Linux couldn't be used without spending thousands of dollars for 64-bit
>> >>> hardware to overcome Linux's poorly designed VFS infrastructure. Windows
>> >>> 2000 was the prime choice. It was the best performing, most stable
>> >>> server software to serve to both the Mac and PC video editing machines.
>> >>> Never failed us once.
>> >>>
>> >>> 2.) My current employer is releasing a product based on EJB. There is
>> >>> very little support, if any from major web application platform vendors.
>> >>> Some provide it, but it's a use-at-your-own-risk type situation. Sun
>> >>> Solaris and Windows 2000 were the platforms of choice.
>> >>>
>> >>> We tried it on Linux, but it performed less than half as well as the
>> >>> Solaris and Windows 2000 implementations.
>> >>>
>> >>> -Chad
>> >
>> >> One more thing I forgot to add...
>> >
>> >> Bottom Line:
>> >
>> >> Linux isn't enterprise ready. It may do static web serving well (not
>> >> the best, but well and cheap) but it doesn't cut it for doing big-boy
>> >> tasks.
>> >
>> >Let me know when you can run w2k on a 244 node S/390 cluster.
>>
>> Would www.bochs.com work? :-)
> To (.):
> Let me know when Linux even appears on any major industry benchmarks
> or shows itself beating the heavyweights in any of the major enterprise
> arenas.
Plainly:
Cost of a 4 machine, quad processor W2K cluster, including licensing for
all software nessesary to serve websites that contain lots of fun doohickeys
like "cold fusion":
180,000 dollars
Cost of 4 machine, quad processor linux cluster that does the same thing:
110,000 dollars.
Cost, per year, of the 3 NT admins who run the W2K cluster:
240,000 dollars.
Cost, per year, of the 1 linux admin who runs the linux cluster:
60,000 dollars.
Number of high traffic, complex, full of frills websites that the W2K
cluster can support:
1900.
Number of high traffic, complex, full of frills websites that the linux
cluster can support:
24,000.
Yearly revenue generated by 1900 W2K sites:
3.42 million dollars
Yearly revenue generated by 24,000 linux sites: (you can charge less per site,
per month)
28 million dollars.
This is based on personal experience and the experience of 5 collegues
in the field.
Do the math. With what todays market is doing (which is dying a slow,
horrible death), the choice is quite clear.
=====.
------------------------------
From: sfcybear <[EMAIL PROTECTED]>
Subject: Re: Benchmark tests - who cares?
Date: Mon, 15 Jan 2001 14:35:02 GMT
I don't see a lot of people in this news group putting down FreeBSD. In
fact, I have seen more posts from Linux users in support of FreeBSD. It
is another OS choice. Just because Linux is the current favorite from
the open source comunity does not put down any other project. Infact, I
think that FreeBSD has been helped a great deal because of the
popularity of Lunux.
In article <[EMAIL PROTECTED]>,
Donn Miller <[EMAIL PROTECTED]> wrote:
> All the bad benchmark test results in the world couldn't make me
dislike my
> OS. Basically, Linux users are kind of spoiled, because their OS is
so much
> better than Windows when it comes to stability, security, and
performance on
> the specweb.
>
> OK, so I run FreeBSD. But, it is the open source cousin to Linux, so
the
> same principle applies equally to FreeBSD and Linux. There isn't a
benchmark
> test result in the world that could make me not love FreeBSD. I think
we all
> remember a time when Linux wasn't all that great, maybe 5, 6, 7 years
ago.
> But did that stop people from loving Linux? Hell no!
>
> Basically, I'm wondering if some people in here are "bandwagon" Linux
fans.
> You know, they like it just because it performs well on benchmarks.
If
> FreeBSD scored 100 worse than Windows on a webbench result, I would
keep on
> lovin' my OS! It's because I know that it is an open source product,
my
> OS would have a tremendous future ahead of it.
>
> All you see references to in here are benchmarks, as if it were the
ultimate
> trait for judging an OS. But, it was *love* that made open source
great ($1
> to John Lennon). Open source OSes (FreeBSD and Linux) overtook and
conquered
> Microsoft because the Linux community was based on love. Micros**t,
OTOH,
> was based on avarice and hate.
>
> So, I say, the hell with benchmarks! Repeat after me: "I love my OS
no
> matter how good or bad it performs on a stupid benchmark!" It is then
that
> you know you are a good advocate, not a bandwagon advocate!
>
> OK, so I admit my post was stupid. But, I just had to react to some
of these
> people in here who base their love for Linux on benchmark superiority,
and
> the fact that 2.4.0 is the best kernel ever. Just love Linux for
being
> Linux. With true OS love, nothing else matters. I love FreeBSD. I'm
sure
> y'all love Linux in the same manner!
>
> -----= 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! =-----
>
Sent via Deja.com
http://www.deja.com/
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Linux is crude and inconsistant.
Reply-To: [EMAIL PROTECTED]
Date: Mon, 15 Jan 2001 15:05:39 GMT
On Sun, 14 Jan 2001 23:36:01 -0500, Gary Hallock
<[EMAIL PROTECTED]> wrote:
>The sort of contradicts the argument that Linux has no quality
>software. DB2 runs on Linux. Gee, you walked right into that
>one!
>
>Gary
DB2 is just what I need running on my home system.
Flatfish
Why do they call it a flatfish?
Remove the ++++ to reply.
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: One case where Linux has the edge
Reply-To: [EMAIL PROTECTED]
Date: Mon, 15 Jan 2001 15:09:19 GMT
On Sun, 14 Jan 2001 21:52:30 -0800, [EMAIL PROTECTED] (Jim
Richardson) wrote:
>Use pan, or I think Knode (or is it knews?). Course, it's a trivial matter for
>slrn+leafnode also, but pan is about as simple as it gets.
pan would be nice if it worked.
leafnode+slrn is an abortion of a way to read news offline. It doesn't
even expire articles with consistancy (read the readme with Leafnode)
and it's difficult to back up retrieve articles you might have missed,
change properties of groups on the fly and so forth.
>maybe it's a multidisk drive?
I suppose.
Still it's not high on my list. I would rather have a decent browser.
>
>like apache? emacs and vim? the wealth of databases?
Typical home applications.
> Maybe the apps *you* use are not available on linux, but the ones I do are,
>and they are a damn sight easier to use from linux than they are from windows
>(if they are even available on windows without mortgaging your house)
To each his own.
>>
>>>That's what it looks like Linux Mandrake is doing. Time to change distro.
>>>Let's see... how much does a CD cost from my favourite emporium? £5? How
>>>much does Windows 2000 Professional cost? £300? What am I to do? Line the
>>>pockets of that company caught being naughty or get hold of free software.
>>
>>It depends on your particular needs. For me, $125.00 or so is well
>>worth it in time and applications. Don't get me wrong I loathe MS, but
>>they happen to make the best OS for the things I need to do and 90
>>percent of the rest of the world seems to agree, at least for the
>>desktop.
>
>great, so use it. What's the big deal?
>
>>>> I see Linux as a crude compromise between cost and time.
>>>
>>>I see Linux as cheap and fun. When it's not being frustrating.
>>
>>It's cheap, and it is interesting and someday it WILL be a major
>>contender for the desktop.
>
>for many desktops, it is now.
>
>>I'm just not certain if I will be alive by the time that happens
>>though, I am 40yo BTW.
>
>you want to off yourself, that's pretty lame, and a poor solution to your
>problems.
>
>
>>Linux is a novelty that ends up folks systems because of curiosity and
>>ends up in the trash can just as quickly because it does not satisfy
>>the needs of the typical desktop user.
>>I have seen it happen 100 times or more.
>>
>
>sure, Nasa is only using it for beowulf clusters 'cause W2k was so capable it
>was boring...
Flatfish
Why do they call it a flatfish?
Remove the ++++ to reply.
------------------------------
** 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 by posting to comp.os.linux.advocacy.
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
******************************
