See also: last week's Back page page.

Linux links of the week

Why be interested in software modems? Have a look at Russ Nelson's linmodems.org site. Russ is working on developing Linux support for the various software "winmodems" out there, and includes a list of reasons as to why that might be a good thing to have.

If, instead, you would like to talk with your Linux system, have a look at LinuxTelephony. This one is a Linux news site with a strong emphasis on telephony and communications issues.

Section Editor: Jon Corbet

Letters to the editor

Date: Thu, 29 Jul 1999 14:41:24 +0200
From: Andrew McGill <andrewm@datrix.co.za>
To: letters@lwn.net
Subject: Software announcements (quirk issues)

Walt Smith make an excellent point about software announcements not
explaining products.  lwn.net has been eulogising zope for quite a
while.  Despite having visited the zope site and having read how good it
is, I still don't know WHAT it is.  Anybody care to explain? (in one
line or less).  &:-)

Date: Thu, 29 Jul 1999 17:39:01 +0200
From: Ul f Dambacher <ulf.dambacher@mach.uni-karlsruhe.de>
To: letters@lwn.net
Subject: LWN990727: Blue Screen of Death

Dear lwn authors, dear linux users, dear linux fanatics

Following linux competing against windows in nearly every situation, I here
raise the question if we shall do compete on this seriously interesting
topic, too.

On the M$ sites you see a highly configurable BSOD filling the whole screen
and giving useful information at infrequent time intervals, on linux you
only get a kernel panic message filling only one single line of the screen,
colored in black and white and only on few occasions.

Where is the kernel hacker writing a patch to configure the kernel panic
messages to show up on timed basis (or depending on system load or user
interaction frequency), giving a little picture and some debug
information (maybe the last two tasks running) and some useful words of
advice what to do next.

Naturally one should be able to configure color and picture and text
contents (I suggest via the proc-filesystem, as modules are not suitable
in this special task).

I welcome any serious aproach,
yours faithfully
    Ulf Dambacher
--
Ulf.Dambacher@mach.uni-karlsruhe.de

Date: Mon, 2 Aug 1999 11:39:44 -0400
From: "Jay R. Ashworth" <jra@baylink.com>
To: editor@lwn.net
Cc: kevin@suberic.net
Subject: Kevin Lyda piece in OSO

In his piece in OSOpinion, Kevin Lyda takes the Linux reporting
community -- comprising such sites as Linux Weekly News, Slashdot, and
Linux Today -- to task based on his perception of their reporting as
"suboptimal".

Now, leaving out for the moment the fact that Slashdot isn't really a
_news_ site to begin with, its name notwithstanding, let us first speak
to his primary complaint.

E*Trade, he says "has reportedly changed it's policy with this share
offer since that time."  He links to a posting at Slashdot... which, as
far as I can see, says nothing of the sort.

The problem here, for those who weren't following the matter, is that
RedHat attempted to make a 'Friends and Family' stock offering to a
small group of notable Linux developers.  Apparently, as far as anyone
can tell, E*Trade failed to properly understand that this was the
posture RedHat wanted them to adopt with respect to this group of
people, and E*Trade instead treated them as normal investors --
subjecting them to an "eligibility profile", which, if the
"E(star)Trouble" site is to be believed, the vast majority failed.

This site, at http://www.concentric.net/~mrsam/etrouble/ and linked
earlier this weekend from LWN, is the saga of one such developer, who
with a $120K income and $60K liquid cash, did _not_ pass their test.

The whole point, of course, is that given RH's desire to make this an
F&F offer, no eligibility guidelines are really applicable; the people
climbing on board for this are _not_ general investors in the vast majority
of cases.

The worst part is that RH is in a very bad position to even apologize
for it, because of the SEC's 'quiet period' regulations.

But, coming back to the thesis of this note, Mr. Lyda's assertions that
the reportage on the matter was inaccurate "because E*Trade has changed
their mind" appears to be itself inaccurate: I can find no evidence
that E*Trade has reversed itself on the eligibility issue (except for
implying that potential participants should lie), nor that they've
restructured the offer in the Friends and Family fashion RedHat
intended.

So, Mr. Lyda; how was this reportage inaccurate again?

Cheers,
-- jra
-- 
Jay R. Ashworth                                                jra@baylink.com
Member of the Technical Staff     Buy copies of The New Hackers Dictionary.
The Suncoast Freenet            Give them to all your friends.
Tampa Bay, Florida     http://www.ccil.org/jargon/             +1 813 790 7592

Date: Tue, 03 Aug 1999 01:36:08 -0400
From: Joe Drew <hoserhead@bigfoot.com>
To: metcalfe@idg.net, letters@lwn.net
Subject: FSF freeloaders?

In response to your July 26 column (
http://www.infoworld.com/cgi-bin/displayNew.pl?/metcalfe/990726bm.htm ):

Ok, it's increasingly obvious you're hostile to Linux. I don't
particularly understand why, but that's not the point of this e-mail.
What I really want to know - and I really, truly want to know this - is
how you can possibly call the FSF a gang of freeloaders. Saying
something like that puts the whole fiasco into perspective: you truly
have no idea about that which you're writing.

In case you weren't aware, and it's highly possible you weren't, Richard
Stallman created the GNU project when he was finally fed up with the
proprietary nature of most software, with one goal: to create the first
fully Free (in both senses of the word; market-types might call it Open
Source) operating system, GNU. Shortly aftewards, in 1985 or so, he
created the Free Software Foundation, whose stated goal was to create
GNU. They would pay people to write it. It's from the FSF that such
works as GNU Emacs, gcc, and lesser-known projects like gzip, make,
autoconf, and the myriad of other tools which make up the basis of a
GNU/Linux system. The Free Software Foundation is the reason most
distributions exist; if anything, everyone else should be called a
freeloader off of /their/ hard work.

I truly would like to know how you can justify calling the FSF
freeloaders. If it was a mistake, correct it - because the Free Software
Foundation, along with all the programmers who wrote code for it, are
the reason Linux is in the limelight today.

-- 

Joe Drew
http://www.woot.net

Don't watch your back - you'll never see me coming anyways.

Date: Thu, 5 Aug 1999 00:16:11 +0400 (MSD)
From: Khimenko Victor <khim@sch57.msk.ru>
To: letters@lwn.net
Subject: Linux fragmentation ? It WILL happen :-/

So far I was optimistic on subject of Linux fragmentation. Since source is
open and all changes can be incorporated in mainstream version without big
effort then what's the need for proprietary branches ? Except of cases where
maintainers of mainstream version are constantly reject much-demanded patches,
of course...

Recently I downloaded libgcj and now I' MUCH less optimistic. Reason ? Ok,
libgcj uses Boehm's GC. Version from http://reality.sgi.com/boehm/gc.html was
not "good enough" for them and so version from libgcj is different. Just how
different ? Well, comparing with gc 4.14 there are one two simple functions
added (GC_debug_object_start and GC_debug_generic_malloc), support for Linux
Alpha & Linux Sparc, support for QUICK_THREADS, autoconf and shared libraries
support but there are no support for for Linux/M68K, no support for Watcom C,
no support for SMALL_CONFIG and no new revision of gc_alloc for SGI STL
versions > 3.0... Nothing spectacular in both versions and merge can be done in
day or so (without full testing of course): I was able to use SGI's version
with libgcj after less then hour of tweaking since I'm not need suppot for
neither Alpha nor Sparc nor M68K. But still we have two INCOMPATIBLE branches
of Boehm's GC. One from Cygnus (even if it's internal version for libgcj 
users of Linux/M68K will suffer, for example) and one from SGI. BOTH companies
are very supportive in regard of Linux (Cygnus -- for long time, SGI -- vocally
only recently) but still we have one small (yet sophisticated enough) library
fragmented for no justified reason. And branch was created by company with
LONG relation with free software community (Cygnus) what makes situation even
worse :-/

If we can not keep ONE SMALL LIBRARY from fragmentation then how we can even
hope to keep full linux suite from fragmentation ? As we can see there are
MUCH LESS incentive to keep unfragmented codebase when corporations are
involved.

And I'm not sure now if Linux community we'll be able to keep Linux
unfragmented under pressure from "big boys" in corporations :-(( Not at all.

P.S. If you have some idea about why it's not common trend but rare exception
I'll be happy. So far I can not find such reason...

Date: Tue, 03 Aug 1999 12:17:05 -0700
From: Anand Srivastva <anand@nmi.stpn.soft.net>
To: bob_o'donnell@infoworld.com, letters@lwn.net
Subject: Bob O'Donnell makes a few startling -- and final -- predictions

Hi,

I don't agree with you on the fate that you have prescribed for linux. I
don't know why people forget how MS became big in the first place. Why
IBM could do nothing against MS, even though MS wasn't bigger than a fly
compared to them. They are still bigger in the revenue sense than MS.
Why could MS win. It won because of two things, IBM-PC was open, and
PC-DOS was also open. Anybody could make a compatible, and write
software for it. The compatibles drove down the prices, and lots of
developers created software for DOS. That's how Lotus, Borland came
about.

But now MS is in the wrong position. Other hardware vendors are becoming
open, Alpha and SGI and IBM. Linux is as open as anything could ever be.
MS can't compete because it is a huge behemoth, gone are the days when
it could move fast. People talk about the turn around in its internet
strategies, but they forget that it has not made even a little money on
that front. It has only made IE free, which is money down the drain,
because there is a deadlock with Netscape. It could only succeed as
compared to Netscape till the time that Netscape was not free. But now
their IE investment is gone. MSN is also not working.

People should start understanding that Internet is a very different
beast, Open Source is a revolution. These two things are actually
different faces of the same coin. And they are akin to the discovery of
the printing press. Open Source movement has been at the center of an
explosion. And it will keep on growing at the same exponential rate as
it has been growing till it covers the whole world, for the next ten
years. The factor which is letting them become this big is the same
openness which had allowed MS to win over IBM twenty years ago.

The reason why Open Source will not splinter is GPL. Some people think
that it is a bad thing to restrict open source with GPL, but it is the
shield that protects all open source software from splintering. If you
look at the history of open source software which are covered by GPL,
you will find that very few have splintered, and those have they have a
very definite reasons, and many like the GCC have also had their
splinters get merged back.

GPL avoids splintering by removing the money motive. You cannot make
money on a GPL software other than providing some added value. And you
cannot make money more than the value that you have added. It also makes
life difficult for a splinter group because it has to improve faster
than the central group, otherwise they lose the value add that is being
provided. This makes it easier to just provide patches to the main code
base, than maintain a splinter base. It is still difficult, but is
easier. The best way is to open the code and let it go in the main code
base. The code released is properly integrated when in the main code
base, so that the vendor has to just add their improvements.

I am predicting that rather than split, open source will force all the
proprietory general purpose OSs out of the market. You are already
seeing a couple of them happening now as I am writing. You are seeing
that the Amiga OS is now becoming just a shell over Linux. You are
seeing that SGI is willing to abandon IRIX in favour of Linux. These
things will become more pronounced as maintaining an OS will become
non-profitable, and without any benefits. You can expect Apple to go
this way in the near future and even IBM to do something like this for
OS/2. This is just the beginning. The two companies which will come to
the fold last are MS and Sun. They are very very closely tied to their
OSs, which will make it very difficult for them to come to terms with
reality.

This forcing out of proprietory software will not be just for OSs. It
will slowly encompass all software that is in general use, or is
interesting enough, or is mission critical enough. There are a few
office suites, there are a few graphics tools, and there are a few
database software. These will keep on getting better and better, till
eventually they will get better than every proprietory software. The
open source software has another property of improving overtime with
use.

I am not saying that all software will become opensource, but mostly the
above three types. Software that is not above three will require some
company to do it, and that will be proprietory.

-anand