More Letters to the Editor

These letters were not printed in the magazine and appear here unedited.

business plan

Just a quick question (worth printing to next issue?)

Is there any (good) GNU licensed programs to help the creation of businessplans?

—Jussi Kallioniemi []

Invitation to Join Freedom of Choice Project

TO: Representatives of... OS user groups, consumer advocacy orgs., ISD associations, software vendors and distributors

FROM: James Capone, of the Linux Info Website
Diane Gartner, Co-ordinator of IACT

The Freedom of Choice Project is a co-operative effort between IACT and James Capone, an IACT member as well as devoted user of Linux, who created the entire project at his own Linux website. In the project's first week, over 5000 people had participated in the Freedom of Choice consumer poll. With help from IACT, James Capone now is expanding the poll to reach users of _all_ platforms.

As you know, all computer users certainly are affected by an ongoing problem in the computer market: Microsoft still maintains an exclusive distributorship with PC makers such as Compaq, Dell, Gateway etc.. Those companies pre-install or “bundle” MSFT software on the majority of new PCs we buy. Once the MSFT software is pre-installed, we may decide to delete it and then fight to get a refund, but that approach still won't get to the root of the problem.

The Freedom of Choice project is our grass-roots, long-term solution. By using the Internet as it was designed—to bring together small groups like ours into a larger, stronger and unique network—we're going to defend the fundamental right of consumers everywhere to choose any and all software that is installed on the new computers they buy.

We want to give users of all platforms the chance to _send a direct message_ to the PC makers, to demand that the companies fully respect every consumer's right to choose. But we need your help to do it!

Join us as *Project Associates* of the Freedom of Choice project. In the spirit of teamwork and co-operation, we are asking you to sign your name to this project, and make a strong commitment to helping the world's computer users regardless of platform, background, nationality or expertise.

The Next Step

Please r.s.v.p. to this letter and tell us if you can commit your organization's resources and name to the Freedom of Choice Project. If you cannot help in an active role as a Project Associate, then please at least consider posting links and announcements for us.

Your active participation would include a few, simple volunteer responsibilities:

1. Devote a small amount of your web space to a “Freedom of Choice” page on your group's website, to outline the goals/benefits of the project and to post frequent updates on the project's growth. Just be sure that your page gives equal emphasis to all OSs and to all computer users regardless of platform, background, nationality or expertise.

2. Directly contact your own members + associates, urging them to visit your website's Freedom of Choice page and then to respond to our Freedom of Choice Poll, so that your group's concerns about software choice will reach the PC makers with maximum impact.

3. Help us to publicize the project by using your media, political and IT contacts. Forward the project's official announcements to your contacts, and urge them to support the project either directly or indirectly.

4. Any other resources or ideas that you'd like to offer....

If you choose to join us as a Project Associates, then with your permission we will add your name(s) to this same Invitation when we send new copies to more organizations; your group likewise will be given a spotlight on our own Freedom of Choice web pages, for the public to see.

The Freedom of Choice Poll is at

Diane Gartner,
James Capone,

Great Article

I really enjoyed reading Alessandro Rubini's article “Software Libre and Commercial Viability” in the February issue.

It is often hard to see where things are going in the Linux and Open Source world, and Mr. Rubini's article was very intelligent and enlightening with quite a few insights into what's up.

—Robert Lynch

Re: University of Toronto WearComp Linux Project (Feb 99)

I've used Linux since 0.99, and taken LJ since somewhere around issue 6 or 7. PLEASE PLEASE don't get political on me. The recent article by Dr. Mann filled the first 3.5 pages with his personal libertarian paranoid philosophy as it pertains to the use of intelligent devices in society.

I am happy for him to tell us about his project, how Linux allows him to be “COSHER”, and to describe his applications of Linux. However, I don't (and don't think most LJ subscribers) give a whit about Dr Mann's personal politics. To read the first three pages of the article, I suppose I should start taking apart my telephones, TV's Microwaves, and temperature-sensing shower heads because there may be a hidden videocam inside them.

Keep up the good work. I love your mag.

—Bill Menger,


I've been subscribing to L.J. for a few months now, and I eagerly await each edition.

I think the magazine needs to dedicate more resources toward editing and reviewing the magazine for errors.

As an example, the latest edition's squid article states Netscape's proxy server does not support ICP, which is incorrect.

I think the quality of the magazine could be greatly improved if the editors worked more closely with the authors. As Linux gains in popularity,

I'm hoping the advertising dollars come in to enable you to continue to improve the overall quality (and width!) of each issue.

—A Demanding Fan, Jim Ford,


Neil Parker wrote:

Read Joseph Pranevich's article on Linux 2.2 in Kernel Korner (LJ: December 1998) with interest. He mentions support for a 'subset of the Linux kernel' on 80286 and below machines. On this tack I was wondering if there is anything geared towards making effective use of ageing 386 systems with 4Meg or so of memory. I have a lab full of these all running Windows 3.1 and would like to convert them to Linux if possible. Probably there are many thousands of similar machines worldwide.

No, not that I know of. That's a bit out of bounds for ELKS. (Linux for 8086-80286)

Your best option here is to compile a main Linux kernel, even a 2.2 kernel with the absolute minimum requirements for your hardware. There are also patches floating around for reducing memory requirements even further, but I can't give you a pointer right now. Having done that, you should strip down the a system to the *bare* essentials. No daemons, no loadable modules, nothing that isn't absolutely necessary. You'll also want to recompile your X server (Mono, SVGA, or VGA16?) with the minimum options and no unneeded drivers.

Of course, this will all have to be done on a bigger Linux system, recompilations on a 386 take *forever*

With all of that done, you'll get a working system. You may need to hack the X source a tad to not configure things like unneeded mouse drivers and etc. With any luck, that will fit, but barely. If you run all applications after that remotely, it might not be so bad.

Make sure you have swap space, Linux 2.2 actually runs faster with it because it swaps some unneeded bits out. And I've heard that Linux 2.2 would be faster than Linux 2.0 but you might want to check both. Alternatively, you could look into building it with a Linux 1.0 or 1.2 series kernel, if your hardware would allow it. And building a libc5 (as opposed to glibc) system might save you a tad more room. And finally, make *sure* that you have no static libs and compile everything shared.

I hope this helps.

Joe Pranevich,

letter to editor

At the UW I work with all platforms but for ease of development I've used Linux since .99pl33

While working on a Journal article (for a medical publication) on computer security, I bought a copy of WinNT Magazine (Dec. 1998) and came across some comments about Linux that I think are unfounded, but I would like some expert and authoritative rebuttals.


On page 35, Craig Barth writes

“According to intelectual property lawyers, the Linux licensing agreement binds any developers who produce software using components of the Linux OS (eg. libraries and runtimes) to release the source code for their additions. This will stop mainstream commerical development dead in its tracks”

On page 122, Mark Russinovich writes

“... whereas WinNT and all commercial UNIX implement kernel mode threads, Linux does not.”

“... the schedular cannot preempt the kernel ... Because Linux kernel is onot preemptable, it is not as responsive to high priority processes as other kernels are ...”

“... the Linux kernel is not reentrant, which means that only one processor in a multi mode system can execute kernel code at one time”

“ For the next couple of years, Linux is stuck with being only a valid choice for small uniprocessor servers ...”


If you could respond with some sources, I would send such back to WinNT magazine.


Intel Red-hat for the LAST time

Several months ago I wrote to you expressing my concerns over certain packages being made available just for users of the Red Hat distribution package. When this letter was published in the January issue I though I might get some feed back but I was amazed to see in the Feb. issue a letter by Fred Nance praising Red Hat.

I think you have all missed the point!

Yes, Red Hat and all the other distributions are doing very positive work in promoting LINUX, Without Red Hat several packages would not have been written and we would have definitely have a smaller user base and certainly not a significant force against Micro Soft.

How long will it be before one large company sides with Caldera, one with Suse and one with Red Hat and we all start to make incompatible packages? If we don't start pulling together soon we will destroy everything already built up and MS will be laughing all the way to the bank.

Could Red Hat give me a positive statement that all their packages including Oracle are compatible with other distributions and they don't use their own libraries and why was glibc2 picked up by Red Hat before it was formally released forcing incompatibility issues?

Thank you for your time.

—Bob Weeks

BTS correction: shutting down a Linux machine

I was just reading the “Best Of Tech Support” in the Linux Journal 1999, and was struck by the “Shutting Down” question, and its answer, since I have a habit of “unorthodox”(?) ways of shutting down.

Among other things, I don't particularly like the ctrl-alt-del method of shutdown, mentioned as “the only way... for any user to safely shut down a linux system is to be physically present at the keyboard and press ctrl-alt-del.” I've found that the odd DOS user will sometimes reboot a system because their window is not responding, so I sometimes disable it in inittab. Also, if this is a “headless” machine with the console on a serial line, there may not *be* a console to type ctrl-alt-del on.

Here is a reasonably simple, and relatively secure way of solving the problem (better checking of the arguments might be wiser). Compile the following, called shutme.c and set the permissions as below (with “user” replaced by the GID of the user, or of a login group allowed to use shutdown:

#include <unistd.h>

main(int argc, char **argv)
  int i;

  for (i=0; i<argc-1; ++i)
      argv[i] = argv[i+1];

  argv[argc-1] = 0; /* execv linkes a null-terminated list of args */


-r-sr-s---   1 root     user        4186 Jan 25 01:31 Monty_only/shutme*
For paranoia, I put this in a directory which was chmod 700 and owned by the user in question, too.

Then, assuming that directory is in the user's path, they can run

% shutme -rf now
without being root, on most unix machines I've used, including Debian 2.1, and almost certainly Red Hat. A nice thing about this is that any flags to “shutdown” are available, including the ability to cancel a running shutdown.

Irix has some pickiness about that, and if Red Hat is similar, then there are some other options.

It's easy for any root process to initiate a shutdown by sending INIT the right signal; replacing the above with just

  execv("init 6",0); /* or execv("telinit -t 10 6",0); */
or, if only shutdown is picky, execv(“reboot”) or execv(“halt”) should be just as good. The disadvantage of these is that they don't issue a wall to all the users, but that can be included in the program as well. Running sync just before is traditional, but init should take care of that without a hitch in modern times.

There are a few other options which *should* work, and do for some linux versions:

Make a group of people allowed to run shutdown, say “shutters”, and

# chgrp shutters /sbin/sutdown
# chmod 550 /sbin/shutdown
# chmod +s /sbin/shutdown
and then anyone in “shutters” should be able to run shutdown.

Lastly, if you make an account called shutdown whose UID is 0 (root) and whose shell is a shell script that runs “shutdown -rf now” (or whatever), you can give people that password to that account and they can

% ssh host -l shutdown
% rlogin ssh -l shutdown
% su - shutdown
and so forth.

I have used all of the above on various unix systems, in various states of security and/or partially-crashedness... I was thinking about it because I had to reboot a half-wedged SGI which I didn't have the root password for, but did have root on an NFS server whose disks it was mounting recently...

Anyway, not meaning to be too picky, but with a name like “best of tech support,” I think this answer fell below standards. I did learn both about the “cp --one-file-system” flag from your column this month, so I *do* appreciate it, by the way, and the info that all zip disks come as partition #4 was an interesting confirmation of a trend I've noticed.

anyway, hope this helps, Thomas...

—Mark Montague,

Letter to the Editor ...

On the same day that I received my Linux Journal, I got a PC Connection periodical. It's a standard PC mail-order catalog, but it seems to assume that people are running Windows. Are there similar catalogs that cater to people running Linux/Unix. Like have non-PnP modems, cards that include drivers for other OSes besides Windows, etc., etc.

—Charles Wheeler,

Article Suggestions

Just a quick suggestion for two articles on subjects that are sorely needed in the Linux community in my view.

Some of us users have some old DOS and Windows applications that we still use and there are no Linux alternatives. This makes us having to use a dual boot system.

Now I have had partial success with Wine in using a specialized communications program, but I have some DOS apps I would like to run under emulation that are indeed simple programs and should be amenable to run under Dosemu. I have not had success with Dosemu in 4 months of trying. I have not been able to find a source that simply explains how the program works and the theory behind how to get it running. The readme files, in my opinion, are very rudimentary.

We need some comprehensive articles on how to set up these applications. Now I know Dosemu has the reputation of just being a hack that so many people are using to run games but there are people like me who want to be able to do some useful work with Dos apps, not necessarily games.

I've had some success with Wine with a specialized communication program that I use to access my patient database at the hospital I work for. It is an old Windows 3.1 app that runs good with a little distortion of the fonts. It is still very usable and negates the need for me to have to boot to Windows every time I need it. One neat thing is that it uses the modem under Windows to make a 9600bps connection to the hospital server and under emulation, I do not see a speed hit whatsoever even though it is running through the extra layer of emulation.

I would hope you would see fit to find an appropriate author(s) who could tackle this task and I believe you would be doing a great service to the Linux community in helping to make Linux more mainstream in being able to run Dos/Windows apps in a Linux system.

I think the time is right to carry this out as I feel Linux is reaching “critical mass” and if a person can see that they do not have to give up on being able to run Dos/Windows programs by migrating to Linux, it might only give an extra boost in acquiring more Linux users.

You have my permission to edit this to your standards if you see fit to publish it in the Linux Journal.

Best regards,
—Kurt Savegnago, M.D.,

No more whinings, please.

I read, from time to time, about user friendly configuration tools for Linux. The last time I did read about is the Davis Brians Letter on LJ # 58 (feb. '99, pag. 6 “Linux Installation and the Open Source Process”). I'm sorry, but I totally disagree with the idea that such a tool is essential, nor useful. The Linux community, the real one, may need tools that help to spread the knowledge about Linux to neophytes, but don't need windozian tools at all. Those that need Bill's nightmare simply must stay with windoze. Linux is the evolution of UNIX (even if not the only one) and it is light-years far from MS, both in power and flexibility. People not sufficiently skilled to install Linux from himselves has at least three choices: 1) to buy a pre-installed system; 2) became a more skilled one or 3) stay with their best silly O.S. Everyone is free to get his own way but, please, don't blame others for your incompetence: Linux may be free (or Open Source or whatever you like to call it) but in the real life nobody can have hothing without some personal effort.

—Franco Favento,

Intel and RH

I have read Robert's letter on page 94, LJ Jan 1999.

I definitely agree with you that we don't need the Intel-Hat, or sth else which is another M$ WinDos alike OS.

In the long terms, it maybe comes to the end that there is only one distribution of Linux surviving in the market, but right now, I hope all the exisitng distributions work ahead, as this will bring the benefits to all Linux users.

In fact, all Linux distributions have the same problems: or says, the aims - they need easier way to install/maintain, and more the better GUI-based applications for business.


re: Dear ljeditor. This little poem might amuse your readers....

I am a M$ separatist, hooked now on Linux. The reason I switched is described in the following little poem:

Flamed on the GPS newsgroup (? by Bill Gates), response in rhyme

Somebody said that I lied
That's bullshit, I quickly replied
An honest mistake
Can anyone make
I'll explain, and then you decide

My ThinkPad and Windows were wed
At the factory in the same bed
They shipped 95
And to keep it alive
An upgrade to 98, I ded

98 was a great deal much wurse
I could tell right away it was cursed (some kind of f.king virus)
It choked very hard
When I tried a sound card
The factory defaults I restured

The most outragous abuse
When Netscape I tried to use
Explorer stomped in
And GPF'd me like sin
My computer I had to rebuse

My times not worth much anymore
I'm just an old profes*sor*
With me its my health
Compared to his wealth
His geek thugs are what I deplore

When at the young age of 23
Keats got a bad case of TB
He then wrote some lines
That have help me define
What a loss Gates handed to me

(“When I have thoughts that I may cease to be,
Before my pen has gleaned my teeming brain”)

Better spent my own time would be
Writing for pos-ter-i-te

So, I've tried to put it in rhyme
Think about this in your spare time
And then you will see
What my time's worth to me
To take it away was a crime.

—M David Tilson,

Is Linux getting too commercial?

I have a total of four PC's, two at home and two in my office. I have in both places one PC with an Uncle Bill Gates' system for talking with electrical engineers, and one PC with Linux system for talking with physicists. I have decided to keep these machines separate, since Uncle Bill has constructed a windowing system with an internal search and destroy program. This program is automatically run as soon as another system (e.g. Linux) is detected on the hard disk. Do not bother to tell me how to overcome such programming games, since I so much enjoyed viewing this program on a friends PC. In an unplanned (by the users) civil war program, Windows 95 took out NT.

However, I would like to express concern about Linux. After looking over the shoulder of a student depositing RedHat 5.0 on an office box, I easily enough installed it again on a box at home. Later I bought some RedHat “power tools”. I ignored the warning on the box that I would need RedHat 5.2. Then I had to go out and buy RedHat 5.2. My discomfort is not yet my anger at Uncle Bill for changing the file format in Microsoft Word, thus increasing the price of talking with electrical engineers. But really, ... 5.0 ... 5.2?

I am perfectly willing to pay a reasonable price for packaged software, without using up my students valuable time searching the overcrowded web sites for the free versions. But I do see some bad handwriting on the wall.

Further, when you merely update from 5.0 to 5.2, rather than doing a fresh install, things that used to work can stop working. My postscript printer file thought the printer moved from America to Europe. It changed the internal view of the paper size and forgot to tell me. The menu bars for increasing the console window size no longer show any words, but I can use the menus. I merely have to remember what the menus used to read when they actually had words on them.

I am afraid that when Intel buys into Linux, the commercial outlook will lessen the quality of the system. Can you get ruined by success?

—Allan Widom,

Re: Small Linux Machines (CyberFlex _isn't_ a Linux machine)

Hi Jeff Alami,

Regarding your statement in 32bitsonline article “Small Linux Machines”

“Which system has the bragging rights of being the smallest Linux computer around? My guess is Schlumberger's CyberFlex Open 16K smart card with the Linux kernel. The computer chip on card stores card holder information and even biometric information for secure authentication. Now there's a small Linux machine.”

I think you are mistaken here. According to the documentation on Schlumberger's CyberFlex web site, this is a smartcard that runs Java programs on a Java Virtual Machine under control of a small operating system called GPOS (General Purpose Operating System). The notion that the card runs Linux seems to have started with Marjorie Richardson's claim to that effect in the January 1999 issue of Linux Journal when granting the 1998 Editor's Choice Award to the Cyberflex. I don't know whether she has issued a correction to her statement since I haven't seen the February 1999 issue of LJ. I'll copy her on this message in case others haven't pointed this out to her.

So it looks like the smallest Linux machine is the one developed from off the shelf parts by Stanford University Wearables Lab. Now _there's_ a small machine that's actually running Linux.

—Ronald L Fox,

See the MUSCLE web site for more about Linux and smart cards.

Reply to 8/98 Article!

In an 8/98 letter Mr. M. Leo Cooper states that one can Symbolically link the Netscape cookies file to the NULL device, thus preventing Heinous WebBots access to this info. ( a Most Laudable Goal! )

Mr. Cooper's soultion is thus:

ln -s ~/.netscape/cookies /dev/null
Mr. Cooper has a Great Idea, on My System; however it Fails, for some Subtle reasons. ( ln ) will Fail IF: The Second File EXISTS ! In other words, as Printed, IF /dev/null Exists, then the Link will Fail ! Important: On Most Unix / Linux systems /dev/null, and /dev/zero are Required for Proper Operation, therefore, Likely to have been Created and Exist.

All is Not Lost though, here is how I was able to Accomplish the job:

1) Delete the ~/.netscape/cookies File, copy it to a Backup if you Want the Info, such as: cookies. Sav or cookies. Bak, Then delete cookies!

2) Use this form for the link command:

 ( ln -s /dev/null
~/.netscape/cookies )
This procedure worked Flawlessly for Me, /dev/null is Preserved, and a New @cookies Link is Created under ~/.netscape !

To Test my theory, I logged into a server that I knew was Particularly Nasty about setting cookies.

After the session I viewed cookies using the editor in Midnight Commander ( a Favorite ), and cookies displayed Absolutely Nothing, the Void that we Want that Uninvited WebBot to See !

This procedure is a little more involved than just piping something like:

ls -l /home/cookies > /dev/null
But Well Worth the effort as “Big Brother” is snooping Relentlessly, and people have a Right to be Concerned for their Privacy!

Now, what do we do about [ Pentium III (c) Intel Inc. ] so called “Hardware Cookies” ? As an engineering student and systems programmer I, and some colleagues are discussing it!

—Jim Boedicker,

spelling error(s).

On page 82 of your March 1999 issue, in the “Red Hat LINUX Secrets, Second Edition” review, you have 3 spelling errors, or mistyping. In the first column, around the 4th line down, it says “Linux kernel 2.2.35”, i believe that that is mistyped. I think it should be 2.0.35. Also, in the second column, 17th line down, it says “(2.2.35)” again, and 5 words later, “(2.2.32-34)”. I'm pretty sure Duane Hellums, the author, didnt mean to type these wrong versions. I just thought i would bring it to your attention.

—scott miga

article corrections

Message-ID: <>

I just read the review of “Red Hat LINUX Secrets, Second Edition” on page 82 of the March 1999 Linux Journal, and I'd like to point out a few problems with the article:

(i) Every time it mentions the kernel, it incorrectly refers to the versions as 2.2.x, when they should be 2.0.x.

(ii) It says, “Also helpful would be a loadable module for sound card support to avoid having to manually configure and rebuild the kernel...” I haven't purchased the book, but if it indeed comes with Red Hat 5.1, as the article claims, then the kernel is already preconfigured for sound card use and installs all the modules that Red Hat supports. Red Hat also provides a nice tool, 'sndconfig', to configure your sound card and modify the /etc/conf.modules and /etc/isapnp.conf file (if needed).

(iii) It says, “...Linux is in dire need of an intuitive, commercial-quality, freeware, GUI-based word processor...” While technically not freeware, both Corel's WordPerfect and Star Division's StarOffice are available for free for individual, non-commercial use. KDE also comes with a text editor that provides approximately the same functionality as Wordpad.

—Jeff Bastian,

Freeware vs. mega$ware

Being in the position in my company to “affect the purchase decisions”, I have encountered some very disturbing trends in the way corporate world sees and uses IT equipment. As many of you out there I have been a victim of the “main-stream” systems which means for example that a $3,500,000 installation by Sun known at the company as “mega-server” every once in a while gives messages saying “No more users allowed to log in”. You know - the licensing thing. The database on which the livelyhood of the company depends is something of a joke. In fact that's a bunch of databases designed by different vendors that handle various aspects of production. They can neither be modifyed nor discarded due to the nature of contracts with their vendors and the sheer amount of money invested in them. So I am currently involved in a project of designing a super-database that would join them all in one working mechanism and would provide a usable interface to the whole system.

You would probably say what can be easier all the tools are avilable for any platform one would choose. Not so fast. The company went shopping for yet another vendor to build the database. The final choice was between (no-no, there was no mentioning Linux!) a company that would charge us $20,000 for the whole project and a company that would charge $500,000. (There's no typos in the zeroes, I am talking twenty thousand vs. half a million dollars.) Now try to guess which company won the gig. The half-a-million dollar one. Of course. And I voted for it too. Why? Because the bigger budget means more “petty-cash”, more restaurant invitations, more business trips etc. Besides some people in the company are dependent on the commissions... What kind of commissions do you get off a Linux-based job? $500, at most.

All of this makes me very depressed. There is no way in the world that Linux can make its way into corporations like this one. But something happened recently that gives hope. I saw a demo by Silicon Grafics of their new Intel-based workstations. They come in two flavors: NT and Linux. And SGI offers full industry-standard support for these machines. Which is great, because they are not cheap (forget about free).

What I am trying to say is, to make it into the corporate world we may need not just commercial, but extremely expensive systems. Like $3,500,000 servers.


Regarding Chuck Jackson's letter in the March 1999 issue (in response to Mr. Havlik's letter), arguing that advertising can easily be skipped, will this may be true, skipping the ads always interrupts the flow of reading.

The same holds true for letters and articles which are “ ... continued on page ...”. I don't see any reason why e.g. the letters to the editor aren't on two consecutive pages in the magazine, but instead are separated by almost ninety pages (in the March 1999 issue). I hope LJ will try to avoid this in the future.

I know that most magazines can't do without advertising, however I would think that the main goal of LJ is to convey information about new trends to the linux community - not every ad in LJ matches this criterion.

Personally, I subscribed to LJ because of the informative articles, not because I like the advertising so much, but perhaps there's a way out for all of us:

I'd like to see a separate section packed with ads, e.g. in the second half of the magazine - just like the german c't magazine - which reduces the number of ads in the first half. This has advantages for both groups of readers, the ones who read the articles and the ones who are interested in the ads. Articles would be mainly in the first half and would contain less intrusive advertising (e.g. ads would only precede or follow the articles, not interrupt them).

—Lars Michael,

Issue 59 editorial comments

Let me briefly state that I have been a loyal subscriber since around issue 5 and thoroughly enjoy LJ.

I'm a bit disturbed by your comments to Reilly Burke who wrote the “Red Hat Phenomenon” letter.

“ . . . However, Red Hat does seem to be the most popular distribution available, so they must be doing something right.”

This sounds exactly like the answer I get when I try to convince users of Evil Empire software that there are alternatives which are more robust and feature rich. Reading between the lines, I get the impression that any distribution which gains market share is a good thing and if one dominates, it must be the most technically sound. By this logic, we should all dump Linux and run one of the flavors of Windows. We should also dump Emacs and LaTex and only use MS Office tools. Clearly they own the largest market share and are therefore are doing the most thing right.

A year or two ago I probably would have agreed but we have seen an astounding increase in Linux coverage in general media. Not just technical publications devoted to Unix but a broad range of publications from Byte Magazine and Computer Shopper (both slanted towards MS) to Dr. Dobb's Journal to mainstream non-technical publications. This is mainly due to the announcements by major vendors to support Linux (Sybase, Informix, Corel, etc), and of course Netscape's decision to open the source to their browser.

I personally use Red Hat and have been happy with it. I will say that I would not rule out switching to a different distribution if it seemed to have advantages.

I am also a bit confused as to what Reilly means by “conventional Unix methods”. I have been using Unix for a number of years and have always thought that there were two main Unix flavors; SysV and BSD. Vendor specific Unix implementations are typically based upon one of these base OSs.

I hope this does not sound too harsh, it is not meant to be. I think you and all the staff at LJ have been doing a great job.

—John Basso,

popular does not equal good

Reilly Burke's letter (March 1999) criticized Red Hat and you responded that Red Hat is the most popular distribution so they must be doing something right.

Well, Microsoft must be doing one hell of a lot of things right in that case, but that does not imply that what they produce is of high quality.

Actually, I did not understand Reilly Burke's criticism of Red Hat as 'difficult to install'. Red Hat 5.2 installed for me with zero hassles. Nor his remark about Red Hat having a 'strange implementation' (unclear what that means). As a UNIX user since version 6 in 1975 and a Linux user since 1995, first using Slackware, now using Red Hat, I have always found all UNIXes to be arcane but very sound. Usually, when someone is criticizing a given UNIX or given text editor, or a given whatever, it is because they are used to something else and are not making room for the learning curve.

That being said, we all know that we have a task ahead of us in developing easier install shells both for the Linux OS and for Linux applications, as well as continued work to agree upon a File Hierarchy Standard and Control Panel for GUI-based System Administration.

—Dennis G. Allard,

editors reply

I the March 1999 issue your reply to Reilly Burke is quite assinine:

“Sorry, you will have to ask Red Hat about their policies. I am not in their confidence. However, Red Hat does seem to be the most popular dis- tribution available, so they must be doing something right. —Editor”

Sorry, why do you think people read your journal but to get information. For you to put someone off the way you did Reilly Burke is about the most pitiful reply I have ever seen. It makes you seem too lazy to ask Red Hat about it (what your subscribers expect you to do) and like an ass for answering in a flippant manner.

WindowsXXX seems to be the most popular distribution, actually, they must be doing something right, you betcha, they market very well, however the software they release seems to me and large numbers of others to be nearly unusable because of stability and frustration caused by difficulty of use.

The man has a legitimate question about Red Hat, so that leads me to conclude that you are (as most journals are) only interested in the advertising revenue.

—Gene Imes,