See also: last week's Back page page.

Linux links of the week

Maybe instead you want to take your Linux with you. The folks at linuxce.org are working to get Linux working on handheld computers and PDA systems. It looks like they have set themselves an ambitious task...

Section Editor: Jon Corbet

Letters to the editor

Date: Wed, 4 Aug 1999 23:24:45 -0500
From: Robert Lipe <robertlipe@usa.net>
To: letters@lwn.net
Subject: Monterey "vaporware"?

Ahem.  You folks generally do a really even-handed job at balancing
things.  This statement missed the boat:

        Most still seem to think, however, that the strongest Unix on
        IA-64 will be "Monterey," a vaporware system being developed by
        IBM, SCO, and others. But by the time Monterey is both real and
        stable, who will still be interested?

Monterey is as real as an OS for an unreleased chip can be.  SCO stated
at SCO Forum in 199_7_ that they had their UNIX ported, booted, and
running on Merced simulators.

We know, too, that VA and Intel are working on Linux for IA64.  Is that
any less vapourware?  I haven't seen any announcements of the public
availability of it, either.

Please don't FUD others while getting so annoyed when Linux is FUD'ed.

Thanx,
RJL

To: letters@lwn.net
Date: Thu, 05 Aug 1999 12:18:41 -0700
From: "   " <lkollar@my-Deja.com>
Cc: 
Subject: Who's afraid of fragmentation?

Not me!

Everyone's Linux setup is a little different, sometimes in ways you
would notice if you sat down at someone else's machine. If I wanted to
be silly, I could make the case that there are 20 million different
versions of Linux, one for each system.

Diversity is good. Each distribution can focus on a particular
strength. Let's take the BSDs for example: FreeBSD optimizes their
distribution for '86/Pentium systems, NetBSD aims to run on
everything, and OpenBSD has great security out of the box. But most
applications written for one of them will run on any of them (perhaps
after recompiling).

This holds true for Linux as well. There are floppy-based
distributions for limited hardware or specific uses, there are
distributions that cater to PowerPC-based computers, and so on. But if
you write an application on your Pentium-based RedHat system, and I
compile and run it on my PPC-based MkLinux system, where's the penalty
for fragmentation?

	Larry Kollar (lkollar@my-deja.com)


--== Sent via Deja.com http://www.deja.com/ ==--
Share what you know. Learn what you don't.

From: Mike Henderson <Mike.Henderson@powerco.co.nz>
To: "'letters@lwn.net'" <letters@lwn.net>
Subject: "Your backup is unsafe, but don't worry too much, it's just a Win
Date: Sat, 7 Aug 1999 12:24:28 +1200 

The problem that Robert de Bath refers to and you cover at some length is
actually also a Windows NT "feature"!

I remember reading a article some months ago that commented on exactly this
difficulty in respect of using MS-Backup to restore software on an NT box.
Basically, you can't.  If something bad happens to your "Program Files"
directory, you have to re-install the software that got munged, not restore
it from backup.  I think it was because there are "8.3" file name links in
the Registry which get broken by the restore process, as Robert points out.
Of course, re-installing e.g. Office after such a problem will probably
trash your Registry anyway, so you're back to "re-install Windows".
Sheesh!

Perhaps the Linux VFAT coders should claim that this is a "works as designed
'feature'" because it accurately reflects the way a Microsoft system works!
ROTFL



Mike

To: editor@lwn.net
From: Nathan Myers <ncm@nospam.cantrip.org>
Subject: KDE coming around?

My previous note to LWN may be bearing unanticipated fruit.  
In last week's Gnomish Bi-weekly News report NO#3
(http://www.linuxpower.org/display_item.phtml?id=129), 
Christian Schaller reported:

  Newsitem 5: Even more code sharing between GNOME and KDE?

  There has been some discussion on the GNOME-KDE list lately
  where Karl Nelson, one of the Gtk-- coders, offered KDE the
  use of his libsigc++ Signal framework, originally developed
  for Gtk--, but now having been rewritten to be of use to a
  wider audience. He got positive feedback from Kurth Granroth
  amongst others, but the KDE people needed more time to discuss
  and think it over because of the extra work switching to this
  framework from QT's it would give them and the implications of
  having a KDE version of QT that works differently than Troll
  Tech's.

Replacing the Qt signaling method in KDE with that in libsigc++ might 
eliminate the nasty Qt macros I reported on earlier.  If true, this 
would very good news, because it would put KDE on much firmer technical 
footing.  KDE would become far less susceptible to accidental breakage 
whenever any important kernel or library header file changed, and make 
it more portable to non-Linux systems as well.

I will find out more about this and report in more detail.  Stay tuned!

Date:	Thu, 5 Aug 1999 15:17:42 +0200 (CEST)
From:	Rik van Riel <rik@reseau.nl>
To:	david@chappelassoc.com.campus-party.org
Subject: Re: "Linux, Linux: Enough Already"

[This is an open letter to David Chappell in reaction to:
http://www.entmag.com/displayarticle.asp?ID=72199120502PM]
------

Hi David,

I have read your piece "Linux, Linux: Enough Already" and -
being one of the Linux developers - am rather surprised at
what you have said in your article.


You wrote:
   "I don't doubt the dedication or talent of those volunteers. What I
   do ... doubt is their sanity. ... they're just giving it away.  ... 
   somebody, somewhere, is absorbing the value that's being created. 
   These enthousiastic Linux developers seem in some ways exploited.

As one of those Linux developers (currently in Spain to give a
lecture - yay!) I don't feel I am losing something by sharing the
code I write. In fact, I only feel enriched by the free flow of
(sometimes very good) ideas that's possible in the Open Source
world.

And about the economic value: most companies have started putting
things like "mindshare", "goodwill" and other intrinsics on their
list of valuables - to be carefully protected an enhanced.

And indeed, when you look at the Linux developers, it'll be pretty
hard to find someone who hasn't got an interesting job, lots of
freedom and all the other things valued by the hacker culture. We
didn't get it by coincident, it is largely a result of the mindshare
we receive as a side effect of our Linux hobby.


Further on, you write:
   "These developers, like lots and lots of other people, hate 
   Microsoft"

Of course, I can't speak for the others, but I know that I
don't. In fact, I don't really care about them at all - even
if they manage to push Linux "out of the market" nothing's
lost. I can maintain the system myself and am not at all
dependant on what's happening in Redmond. Although I can't
read their minds, I suspect that this holds true for quite
a number of Linux developers.


And also:
   "[technical superiority is irrelevant], as long as two technologies
   are each good enough to solve the relevant business problems"
   "Who wins depends on support, marketing, availability of applications
   and various other characteristics"

I can accept the point about technical superiority (we could
argue about NT4 not being good enough, but W2K is around the
corner and I really don't know enough about that to make much
sense). Marketing also is a point where Microsoft is winning
hands-down - at the moment -, but that might well change now
that SGI, IBM, HP and various other large companies are supporting
Linux.

And that brings us to support and the availability of business
solutions -- if there's anybody who seriously believes that
IBM, SGI and HP couldn't do that just as well as Microsoft,
could (s)he please raise his/her hand?  In fact, because of a
working market mechanism in the Linux world, support and other
added values will probably be _better_ than with Windows.


Good advice(?):
   "My advice is to use Linux where it makes sense, in places
   where a cheap Unix box is a good fit. ... [organizations]
   should already be Unix-friendly"

Well, obviously you should only use it where it makes sense, but
that's the case for any tool. What I don't necessarily agree with
is that Linux is only suitable for simple and cheap tasks.

Just take a look at the 1400L from SGI - $8000 just for the entry
model in this line of machines, not what I'd call "cheap"...
And the users are already asking for bigger Linux machines than
the ones that are shipping right now, or - like Fermilab - they
are using a huge number of fairly large Linux machines in a cluster,
achieving TOP500 supercomputer performance, not what I'd call "small".

As for the need for competent administrators, you need them for
_any_ system. And of course, the admin in question should be
familiar with the system (s)he is maintaining. If the task isn't
very extreme, just ask the admin which system (s)he's the most
competent with and go with that.


And finally:
   "every installed Linux box is more likely to be a lost sale for
   Sun, HP or IBM than for Microsoft"

I wonder why you didn't include SGI in the list. Or actually, I
have a pretty good idea why you didn't include it...

The above companies are _hardware_ vendors who have no interest in
building and supporting their own Operating System. After all, any
commercial OS is expensive and only adds to the cost of the machine
(driving customers away). By sharing the development expenses with
the Linux community the cost of all high-end servers can come down
quite a bit - attracting customers from the high end _PC_ market
and not from each other!

Let's consider the situation where an organization is running a
service on Intel hardware, but the system demands are simply
outgrowing the hardware capabilities of the architecture.

Now an NT user will be stuck, with nowhere to go. The Linux users,
on the other hand, can simply buy one of the bigger boxes from Sun,
SGI or whomever and continue using Linux. And since Linux always is
Linux, a simple recompile of the server software is enough.

Of course, if the software itself is commercial you won't be able
to do the recompile yourself, but there's no reason to assume that
software companies will refuse to serve a market that's just one
compile away. Looking at the famous "We just typed make" document
from Oracle should be enough to convice you of that.

Personally, I believe Linux will be a big threat to NT. Not
because of personal reasons, but simply because I think that
customers are tired of the "one size fits all" solution
Microsoft has been trying to push upon us.

Users want choice, and with the current "we come in all sizes,
shapes and flavours" situation that the Linux community has
to offer, we can give just that -- and the best part, we don't
even have to sacrifice compatability...

regards,

Rik van Riel.