Linux-Advocacy Digest #552, Volume #26 Wed, 17 May 00 03:13:05 EDT
Contents:
Re: Here is the solution (Leslie Mikesell)
Re: Here is the solution (Christopher Browne)
Re: Things Linux can't do! (Perry Pip)
NYC LOCAL: Wednesday 17 May 2000 NYLUG Meeting: Lars Marowsky-Brée on SuSE and
Highly Available Free Systems ([EMAIL PROTECTED])
news: Intel targets Linux as first OS on ITANIUM (bob@nospam)
Re: Things Linux can't do! (Perry Pip)
Re: What have you done? (Rob S. Wolfram)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Leslie Mikesell)
Crossposted-To:
comp.sys.mac.advocacy,comp.os.ms-windows.nt.advocacy,comp.os.os2.advocacy
Subject: Re: Here is the solution
Date: 17 May 2000 00:54:08 -0500
In article <YhkU4.69718$[EMAIL PROTECTED]>,
Daniel Johnson <[EMAIL PROTECTED]> wrote:
>> By 'application' I assume you mean any product that Microsoft themselves
>> could write.
>
>That clearly includes OS components, for *any* definition of OS
>that includes any part of Windows. After all, Microsoft makes
>*all* of Windows.
Yes, in particular any components that are extra cost, or talk
over the network with a per client/connection fee should be
considered a separate item open for competition. You should
never be forced to buy something for the other end of your
connection just because you have a bundled client for it,
or to use a particular client because only the same vendor's
software will accept your password.
>If you way Microsoft must make it possible to rewrite any
>part of Windows, then you are demanding they open source
>the fool thing, and thereafter not make changes (except additions).
No, I am asking for well documented APIs and protocols.
>> I don't think you can write a domain controller from
>> the existing specs.
>
>But you can. You just can't replace *part* of Microsoft's domain
>controller software, and keep part. If you want to provide your own
>domain controllers, you must provide the client side as well- but
>Windows provides the hooks to do this.
Are you sure (that you actually can, not that they claim you can)?
I was fairly sure I had read about someone's attempt that failed,
but can't recall the detail that was missing. And of course
win2k changes the rules again.
>> I don't think you can do client software that
>> is capable of the 'one signon' trick on NT by transparently handling
>> passwords for multiple services.
>
>You can. Look up "Security Support Interface", if I recall my
>acronym correctly.
This is pretty complicated stuff to wade through. Is there an
instance of anything that is working with it? That is some
non-MS code interoperating with both a domain controller and
something else?
>> I don't think Microsoft should
>> be able to pick and choose what products a competitor is allowed
>> to write that will interoperate correctly.
>
>Why not? They seem *very* open-minded about it. They
>provides hooks for all sorts of things. Does, say, Linux provide
>a similar way to plug your own security authentication system,
>so that the standard security APIs all use it?
Yes, the mechanism is there in the form of PAM (plugable authentication
modules. Most, if not all Linux distributions are using it and
you can authenticate against an NT domain, LDAP, etc. as well as
normal password files and NIS. Solaris uses this approach too.
>If that's a problem then MS's withholding new APIs *isn't*- it takes
>some time for a new API to filter out and become widespread.
It takes exactly as long as MS wants it too.
>> Of course MS can just quit supporting the others any time they
>> want, so it's fine for them. After they develop to the new
>> api they can just push the customers there. Nobody else can
>> consider that option.
>
>They can consider it; but like Microsoft they face the fact that it
>is real hard to get users to switch like that. There is reason why
>Microsoft bends over backwards for backwards compatibility, and
>it is that they are afraid their users will not come along on the upgrade
>ride if they don't.
Backwards compatibility? You mean like allowing hardware vendors
their choice of continuing to bundle a Win 3.x package or ship Win95?
Or providing dial-up TCP for Win3.x that meshed with win-for-workgroups
networking? That would have been bending over backwards. No, that
would have been just reasonable. They didn't do that.
>That argument applies to other developers as well, of course.
No other developer gets to pick the time that the pre-win2k versions
stop getting any support. If history repeats, it will happen
as soon as that apps division has a version of Office that requires
the new API.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: Here is the solution
Reply-To: [EMAIL PROTECTED]
Date: Wed, 17 May 2000 06:05:27 GMT
Centuries ago, Nostradamus foresaw a time when [EMAIL PROTECTED] would say:
>On Wed, 17 May 2000 03:31:08 GMT, [EMAIL PROTECTED] wrote:
>>In article <8fqekl$294r$[EMAIL PROTECTED]>,
>> [EMAIL PROTECTED] (Leslie Mikesell) wrote:
>>
>>> And the starting price for a VMS cluster would be???
>>
>>I doubt I've spent $1,000 on my home VMS cluster, which has about 5 or
>>6 VAX nodes, and 1 Alpha node.
>>
>>The real problem is power. My MicroVAX 3900's have 3 kilowatt power
>>supplies, and it is expensive to run them. Even a modern Alpha will eat
>>up a lot of power as well. But the cost of the machines is minimal (and
>>software of course is free), at least if you get something aside from
>>the newest generation. Unlike Winux and Lindows, VMS runs perfectly
>>well on hardware which is not the latest and greatest
>
>Define "perfectly well". Can it play Q3? Run Photoshop or The Gimp?
>Run at 1280x1024x24 bits without slowdown? Network at standard
>100baseT? Easily dial via PPP with a normal modem? Include normal 16
>bit sound?
>
>A BookPC + Celeron 533 + 8G hard drive + 64M RAM does all this, and
>for about $400.
>
>So, who would bother with VMS? Why?
>
>It's bizarre what some people will put themselves through...
Fighting with sendmail.cf is something else bizarre that some people
put themselves through.
The thing that is fair to say to denigrate VMS is that the only way
to get those machines _cheaply_ is to take cast-offs that _someone
else_ paid dearly for.
The notion that the machines or software is cheap or free is only
true so long as the only people going after these machines are a
small population of "daft tinkerers." (Parallelling the "daft
tinkerers" that want to port Linux to NeXT cubes, Vaxes, and such.
It's not just VMS folk that are daft...)
If 50,000 would-be home users started looking for Vaxen to run VMS
clusters at home, you'd see the prices of used machines move back up.
At Ebay, there appears to be _one_ VAX for sale, that one being
one that was originally priced at around $900K, for use aboard a
military aircraft, bidding at about $500. With marketplaces like
that, if people wanted 'em, the prices would rise. (Probably not to
$900K, admittedly!)
The other thing that I need to comment is that many of the _other_
merits that tsm claims for VMS have at least _some_ degree of truth
to them.
Critically, the world would probably be a rather better place, at
least vis-a-vis computing, if there was a bit more "diversity" of
platforms. While I never particularly _liked_ VMS, I did not by
any means _despise_ it.
The years of the "Microsoft Hegemony" have been about as bad as the
times of the "IBM Hegemony." Unfortunately, in despising Microsoft,
seeing UNIX as the only persistent alternative leaves the risk of
abandoning one tyranny for another. All of the following options
are pretty bad:
a) A world where there are only IBM computers, IBM compilers,
IBM peripherals, ...
b) A world where there are only Microsoft operating systems,
Microsoft applications, Microsoft certifications, ...
c) A world where all OSes are modelled after UNIX.
In between the days of "absolute IBM dominance" and the days of
the "Microsoft Hegemony," there was a brief period, somewhat
parallelling the brief times of German and Russian republics between
their tyrannies under past monarchs and the likely more evil
tyrants, where a whole lot of different systems blossomed.
There was OS/9. TOPS-10 and TOPS-20. GCOS. Almost as many
variations on UNIX as there are Linux distributions today. VMS.
All had some useful ideas to contribute; they had different strengths,
weaknesses, and vulnerabilities.
I _can't_ see VMS making a serious comeback; too many pieces have
flown away, between RDB, David Cutler, and various Digital downsizings
between then and now.
To an extent, tsm is a bit _foolish_ to suggest that VMS should
be reasonably suggested for home use. There are a variety of
reasons why it would be hard to make that work:
- Users have to search for cast-off equipment, which is hardly a
general solution;
- Training, "user-friendly WIMP software," and Helpful Experts
may prove a mite difficult to come by;
- The electrical and cooling bills would be horrendous.
- DCL sucks (I guess that's two "bashes")
That's not to say that those that want to tinker with VMS should be
burned at the stake.
VMS does offer:
- Rather higher reliability than UNIX systems have traditionally
offered, although as work goes into Linux, the urgency of this
difference diminishes;
- A pretty complete online help system for the system documentation;
- A high speed nearly-built-in relational database, RDB.
When they built NT sort-of-based-on-Prism that was
sort-of-based-on-VMS, they messed this part up...
- A rather different way of accessing disks, providing a combination
of names, links, and "physical drive ID" information.
I somewhat prefer the way UNIX emulates Multics in abstracting this
away via mount points, but I'd not be shocked to see that the way
VMS handles it has some merits.
- Handling dates in a way other than "milliseconds since Jan 1 1970"
has Some Merits. I have little love for time_t as it is.
- Files with record formats.
On UNIX, files are just "bags of bytes." You can do anything with
that, but that approach is not necessarily _fast_. VMS allowed
attaching further format information to files which could allow
interesting access methods.
- DECNet was proprietary, but certainly Sucked Less Than NFS.
I like to blame the brain damage of NFS on MS-DOS (they tried to keep
NFS compatible with the semantics of MS-DOS); 'nuff said.
- Different security systems means different non-overlapping
vulnerabilities.
I would find it appalling if, a year from now, EROS, the "OS Eric
Raymond seems to think could succeed Linux" got turned straight into
a Linux user space. It's a _VERY_ different OS, and using it _well_
will likely mandate thinking thoughts that are different from those
that would be thought on UNIX.
--
"It is far from complete, but it should explain enough that you don't
just stare at your sendmail.cf file like a deer staring at an oncoming
truck." -- David Charlap
[EMAIL PROTECTED] - - <http://www.ntlug.org/~cbbrowne/lsf.html>
------------------------------
From: [EMAIL PROTECTED] (Perry Pip)
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Things Linux can't do!
Reply-To: [EMAIL PROTECTED]
Date: Wed, 17 May 2000 06:00:52 GMT
On Wed, 17 May 2000 00:03:34 +0200,
Paul 'Z' Ewande© <[EMAIL PROTECTED]> wrote:
>
>Perry Pip <[EMAIL PROTECTED]> a écrit dans le message :
>[EMAIL PROTECTED]
>
><SNIP> Some stuff </SNIP>
>
>> As soon as you realize all you have proven is that W2K can be used in
>> a Web farm, where stability of single machines is not that much of an
>> issue. You haven't proven or disproven anything in regards to the
>> stability of W2K on single machines under heavy load.
>
>Ok, here's the deal. Define heavy load
You can define heavy load in alot of ways. Basically running at or
near system limits, with occasional demand to go beyond system limits
could be considered heavy load. But there are various limits in the
system. Some imposed by hardware (processing power, memory, i/o
bandwidth, etc.) and others built into the OS (processes,
threads. etc.). So your heavy load may not be the same as my heavy load.
If more demand is placed on a system than the limits will allow, then
the OS should deny resources (deny malloc() calls, fork() calls,
network connections) gracefully and revover cleanly when demand
returns to a sustainable load. A good OS should be able to be brought
to it's knees under extreme load and then be able to recover when load
drops down.
>with documented example,
Huh? I'm not going to document anything. You are the one who wants to
prove something.
>and if I can
>I'll try to find something that matches your definition.
If you have any cases of single W2K machines (not farms) running under
the circumstances above I would like to hear them. A W2K machine may
be stable under one type of heavy load but not others.
>What do you think ?
Or you can call a truce.
Perry
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.os.linux.misc,comp.unix.bsd.freebsd.misc
Subject: NYC LOCAL: Wednesday 17 May 2000 NYLUG Meeting: Lars Marowsky-Brée on SuSE
and Highly Available Free Systems
Date: 17 May 2000 02:10:57 -0400
Though it is now admitted that Free Software is the right software for
pile-o-peecee parallel compute clusters, there is another sort of cluster
which is equally important: The Highly Available Cluster. Such clusters
are used by the military and by large businesses to provide services which
must be available at all times.
LXNY neither confirms nor denies that SuSE is working on a free
MULTICS-like system.
http://www.best.com/~thvv/tvv.html
http://www.multicians.org
http://technocrat.net/958163435
Official NYLUG notice below.
Jay Sulzberger <[EMAIL PROTECTED]>
Corresponding Secretary LXNY
LXNY is New York's Free Computing Organization.
http://www.lxny.org
*** New York Linux Users Group May Meeting ***
- NYLUG.org -
SuSE Presents the Latest in High Availability for Linux
5/17/00
Wednesday
6:30pm-8:00pm
IBM Headquarters Building
590 Madison Avenue at 57th Street
Check in at lobby for badge and room number
Two short years ago, skeptics used to complain, "Linux will never
succeed... There are no large companies behind the platform... There's no
support... There is no top tier relational database for Linux... There is
no journaling filesystem and how can you expect me to deploy it in a
production environment when there is no high availability?" It took some
time, but one by one, each of these criticisms has been bowled over by the
ever-accelerating snowball of Linux.
Based in Nurnberg, Germany, SuSE AG is a global software company that
among other things, produces the well-known Linux distribution by the same
name. In addition, SuSE is pushing the envelope of production-level Linux
that is appropriate for use by banks, trading houses and other entities
with IT infrastructures that simply can not go down.
High availability enables you to link two or more servers together so that
one transparently picks up the computing load should the other fail.
SuSE's HA solution removes any single point of failure and allows
applications to increase availability to the level required for
mission-critical data center operations. This failover capability is an
important component of the maturation of Linux as complete enterprise
computing environment.
Recognized for his work in the Linux FailSafe, Linux Virtual Server, Linux
HA, and Heartbeart projects, Lars Marowsky-Brée will answer your questions
and showcase the cutting edge of high availability Linux. This is a very
technical meeting, but should not be missed by either newbie or elitist.
_______________________________
After the Meeting... Stammtisch
Join us around 8:15pm or so at the Typhoon Brewery & Restaurant located
at 22 East 54th Street between Madsion and 5th Aves.
What is Stammtisch? http://linuxmafia.com/bale/#stammtisch
______________
Swag Give-Away
DeCSS t-shirts, more DeCSS t-shirts and even more DeCSS t-shirts. Wear it
in the State of New York and you risk getting arrested. Better to wear it
Washington DC where free speech rights are actually protected. Thanks to
Steve Blood for providing us with this cool swag and many thanks to
everyone who demonstrated against the DMCA earlier this month. -- Stay
tuned for an upcoming "official" article from NYLUG.org --
______________________________________
Where & When is the NYLUG.org Meeting?
With the generous support of IBM, all regular NY-LUG meetings are held at
the main IBM building at 590 Madison Avenue at 57th Street in mid-town
Manhattan every third Wednesday of the month starting at 6:30pm. All
meetings are free and open to the public.
_____________
Mailing Lists
* nylug-talk
Talk Mailing List for people to contribute and resolve each other's
technical problems.
* nylug-announce
Announce Mailing List for speaker and installfest announcements.
To post a message to the talk mailing list, address it to
[EMAIL PROTECTED]
NOTICE! The flamer problem has been solved. Personal attacks are now handled
offline and not on this public mailing list. In fact, flamers are
immediately unsubscribed without warning. If you don't see your post or
anyone else's on the mailing list, you'll know why.
_____________
List Archives
Thanks to Eric Berg, we now have sortable archives from 1999 and in new
millenium. Check it out at: www.nylug.org
________
Contacts
Jim Gleason, President, [EMAIL PROTECTED]
Eric Berg, Vice President, [EMAIL PROTECTED]
______________
What is Linux?
http://www.nylug.org/about_linux.html
__________________
What is NYLUG.org?
NYLUG (www.nylug.org) is New York's Linux Users Group supporting all
things Linux and Open Source in the New York metro area. Please feel
free to contact me if you have any questions about the NYLUG.org.
______________
Special Thanks
Many thanks go to all of the people who travelled down to Washington DC for the DMCA
protest in front of The Library of Congress on May 2nd. Many thanks also to Rebecca
Blake & Lara Kisielewska of Optimum Design (they co-publish Linux Magazine) for
creating
our mascot "Tuxi" and to Mark Andal for the original cabbie idea. Thanks to Ari Jort -
core nylugger whose contributions are too numerous to mention here. Special thanks
also
go to Peter Norton and Barry Hughes, NYLUG's lead engineers at LinuxWorld Expo, and
all
of the other NYLUGGERS who volunteered. Without you, we could not have done the show.
Good job!
_______________________________________________________________
NYLUG.org: Host LUG Sponsor of LinuxWorld Expo's .Org Pavillion
February 1-4, 2000; Jacob Javits Center, NYC
===============================================================
Jim Gleason VA Linux Systems
email: [EMAIL PROTECTED] http://www.valinux.com
phone: 212-858-7684 President, New York Linux Users Group
fax: 212-858-7685 http://www.nylug.org
===============================================================
------------------------------
From: bob@nospam
Subject: news: Intel targets Linux as first OS on ITANIUM
Date: 16 May 2000 22:29:06 -0700
INTEL PUTS LINUX FIRST ON ITANIUM
According to undisclosed sources inside Intel, Linux is now the
number one target platform for Itanium, the chip giant's new 64-bit
CPU. Why else would the company have availed Linux developers of so
much sensitive technical information about the chip? ZDNet offers
no less than three stories on what could be an historic shift:
http://cgi.zdnet.com/slink?34966:9737542
http://cgi.zdnet.com/slink?34967:9737542
http://cgi.zdnet.com/slink?34968:9737542
------------------------------
From: [EMAIL PROTECTED] (Perry Pip)
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Things Linux can't do!
Reply-To: [EMAIL PROTECTED]
Date: Wed, 17 May 2000 06:28:25 GMT
On 17 May 2000 05:54:04 GMT, Stephen S. Edwards II
<[EMAIL PROTECTED]> wrote:
>Perry Pip <[EMAIL PROTECTED]> writes:
>
>: Stephen, You have already shown yourself to be a liar:
>
>: http://x46.deja.com/getdoc.xp?AN=624137505
>
>Your reply was not worth responding to. I stated very clearly that the
>majority of the problems I've ever witnessed, or experienced with
>WindowsNT were fixed by changing out hardware. As for my own problems
>that I encountered, yes, _all_ of them were attributed to hardware.
That's not what you said. I think you're a liar. Let people read your
posts and judge for themselves:
http://x46.deja.com/getdoc.xp?AN=623637730
http://x46.deja.com/getdoc.xp?AN=623940112
>Your lack of English comprehension validates no reasons to call me a liar.
Now you're stooping to insults.
>I find that you argue much like Charlie. You offer nothing but anecdotal
>rambling, and you never seem to offer hard factual data to prove your
>claims, or the claims of others which you seem to agree with.
What claims?? Provide references to posts of whatever claims I have
made. All I have done is convey experiences of myself and those around me.
>: You have also shown yourself to be somewhat unreasonably prejudicial:
>
>: http://x46.deja.com/getdoc.xp?AN=624188730
>
>Perhaps. I can easily say the exact same thing about you.
Show me where. Give a reference to one of my posts were I act in such
an arrogant way as you act towards all users of an OS.
>: I never said Charlie proved anything. I don't know where you got that
>: idea. All I said was that Paul hasn't proved anything either.
>
>I got that idea from the fact that you've been supporting everything he's
>been saying.
Show me where. Give a reference to one of my posts.
Perry
------------------------------
From: [EMAIL PROTECTED] (Rob S. Wolfram)
Subject: Re: What have you done?
Date: 17 May 2000 06:20:48 GMT
Reply-To: [EMAIL PROTECTED]
Full Name <[EMAIL PROTECTED]> wrote:
>You want to secure an NT file/print server? Easy. Delete the TCP/IP
>protocol and run a non-routable protocol such as IPX. To achieve the
>same level of security with a Unix box you would need to spend a week
>wrapping all the TCP ports.
ROTFLOL!
Honestly, while reading the first part of your post I actually thought
that you were actually knowing what you were talking about... until you
advise IPX as a secure *non-routable* protocol! LOL!
Or was it a typo? Did you perhaps mean NETBEUI?
>Unix's ugo - rwx permissions are simply inadequate for a modern
>computing environment.
Can you examplify this? ACLs are nice and available to Unix and Linux if
you really need it (Posix ACLs), but they imply more administration if
you don't have sane defaults (that ugo provides) and by their very
nature more complex to implement, while ugo +sticky bits and suid/sgid
bits are just part of the Unix inodes.
Do some homework before you make a fool of yourself again.
Cheers,
Rob
--
Rob S. Wolfram <[EMAIL PROTECTED]> PGP 0x07606049 GPG 0xD61A655D
Beauty is irrelevant.
-- Seven of Nine, stardate 51186.2
------------------------------
** 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
******************************
