LJ Archive

More Letters

Re: April 2000 Issue, Number 72, Letters column, “In Linux We Trust”

Laurie Dare's letter on the Linux Open Source Expo and Conference Program in Sydney (Letters, April 2000) struck a chord with me, and I decided to answer the letter.

First of all, I will mention that this was my third trip to Australia in the past twelve months, where I have given talks at much lower-priced venues, such as CALU and AUUG, as well as free talks at various user groups in Australia.

The $1350 AUS price quoted by Laurie in the letter was the price for both the Linux conference and the ASPCON conference. The Linux conference by itself was $695 AUS, with a Government/Academic discount of 10% off that OR an active LUG member being able to get in for $595 AUS. The dinner price was correct as stated at $125, but again I would like to point out for our non-Australian readers that this was $125 AUS, and not 125 USD. Since USD were trading for 1.82 of AUSD that day, it means the dinner really cost about 68 USD, not an unreasonable price for a dinner in Sydney.

Nevertheless, I was a bit upset to hear that the dinner price was even that high, and I spoke to the organizers about it. They assured me that the dinner was a “break-even” event. I can imagine that, since the Sydney Linux User's Group tried to rent a room for a four-hour meeting in the same facility, and were charged $500 AUS (274 USD) for the privilege of having a room and chairs. This $500 AUS charge was picked up by LinuxCare, much to the relief of the SLUG.

Still not completely mollified by the explanation, I proceeded to put on five more one-hour talks over the next three days of the exhibition, at the small theaters in the Red Hat, Linux Care and SuSE booths. These talks were free of charge, and were well received by the people who attended. Other “luminaries” such as Bob Young, Dr. Andrew Tridgell, Paul “Rusty” Russell and others talked in these theaters as well.

I believe that the Linux community will have to realize that as Linux events get bigger the facilities that can handle them get fewer. You can not have as visible a presentation to 2000 people in a room that has eight-foot ceilings the way you can with a room that has twenty-foot ceilings. And while you have to have large spaces for keynotes such as Linus' keynotes (which tend to draw 6000 or more people), you also need to have smaller rooms for sessions. These types of facilities are hard to find, expensive to build and maintain, and expensive to rent. Nor can an overhead projector or inexpensive LCD projector do the job of a professional LCD projector (which can cost $500. USD per DAY to rent). Larger events in public places mean the forced use of paid union personnel rather than volunteers.

The exhibition space that was chosen (Darling Harbor's Sydney Exhibition Center) is probably one of the most expensive in Australia. On the other hand, it's proximity to one of the largest population and commerce centers in Australia allowed for a huge number of people to see and find out about Linux. The number of “newbies” that stopped in the exhibition to find out about our operating system was exhilarating.

Still, Laurie has a point, and we should address it. In the beginning of the Linux movement, there were only a few people interested in it enough to go to a conference/show, and those people could fit in one, two or three conferences around the world. “All” of the people were technical, and most of these people were happy to have a place to sleep on a couch in an organizers home (Hi Donnie!).

As the Linux interest grew, more and more events started to occur, with more and more people going to each event. These events not only grew in size, but took on three different flavors for three different markets. There is a fourth “market” that will be discussed later.

The three main markets for “Linux Events” are: the 'Introduction to Linux Newbie Exhibitions', the 'Showcase to the Linux Faithful Exhibitions', and the 'Hackers Delight Conferences'.

The 'Introduction to Newbies“ event should be an exhibit only, with a few talks given to educate. The event is paid for by vendors with exhibit space. Entrance is free to visitors. This is typically located next to a Micro$oft Windows event, or an ISP event, or some other event that will attract the Linux illiterate. Hard Core Linux people may want to stay away (unless you wish to help staff a .ORG stand). The Linux Business Expo (associated with Comdex) is a good example of this type of event.

The 'Showcase to the Linux Faithful' event should be mostly paid by vendors, with enough charge to the conference to cover costs of the conference. Newbies and programmers and sys admins who are paid to go to the event by their management are the main attendees, with a deep discount for students and LUG members. Hard Core Linux people will go to get free CDs and T-shirts, and to attend RAVE parties. IDG's Linuxworld is a good example of this type of event.

The 'Hackers Delight' conference is where the venue is a university, seminary or some hotel in off-season (skiing area in summer, beach in winter, summer camp in late fall, etc.) to house the conference. Attendees stay in local homes of other hackers, or in dormitories. Vendors are limited to 10x10 booths (o.k., for some 10x20, but all simple booths, not huge), and their money goes into funding speaker travel. Realize, however, that volume and quality of attendees are still typically necessary to attract the speakers and vendors attendees would want, so while these events may still be all-volunteer driven, you may find some of them using paid organizers to help with logistics. CALU and the Atlanta Linux Showcase are good examples of this last type of event.

The last ”market“ for Linux people is not a market at all, but will evolve over the next few years, that of the Linux Workshop. Attendance by invitation only, the list compiled by Linux developers. The venue and travel will be funded by vendors. It will be a place where the developers meet to solve hard problems, or to argue at high bandwidth serious decisions. Perhaps this will be co-located with the ”Hackers Delight“ events, but it will also be separate so the developers can concentrate on the issues instead of what talk they are going to give in a day or two.

These markets have not yet firmed up. You will still find some cross-venues as the people who put them on struggle with the definition of their audience. However, with 30 years in the computer business and many, many trade shows and conferences under (and over) my belt, I have a feeling that it will work out this way. And although I have been known to wear a suit and may present at some of the ”newbie“ conferences, I have a feeling that both Laurie and I will feel more comfortable at the 'Hackers Delight' conference.

Warmest regards,

—Jon 'maddog' Hall, Executive Director, Linux International, maddog@valinux.com

Letter to the Editor

I would like to respond to Phil Hughes' comments on Perl in “Watch out for the Snakes” (January 2000).

The Perl motto of “There's more than one way to do it”, while acknowledging the existence of the quick-and-dirty solution, certainly doesn't encourage it. Efficient programming in Perl comes with practice and effort as it does in every language. But very often the quick-and-dirty solution will work in Perl due to the forgiving nature of its interpretor.

Object-oriented programming in Perl is achieved through the use of modules. It is usual to include information about object variables, instantiation and methods in a separate module file. The definitions of objects in this manner is remarkably easy, as is their use within a program. The archive site http://www.cpan.org/ has over a thousand such modules available for download, testimony to Perl's re-usability and object-oriented capabilities.

Finally, I have been an avid Perl programmer for over two years, was able to write working programs in a short while, and have since been kept interested by the breadth of the learning curve. I have never used sed or awk, only wrote my first shell script a month ago, and thoroughly recommend Perl to anybody.

I have no doubt Python is a useful language in its own manner, and would appreciate hearing more about its features on their own merit, rather than at the expense of Perl. After all, this isn't an election campaign.

—Christopher Dawson, cmd@maths.uq.edu.au

Keep up the good work!

I have to admit it took a while before I found all your excellent writings, Cluetrain and whatnot. Actually, it was your “Linux for Suits: The Real Meaning of Markets” in the February issue of Linux Journal that I stumbled over yesterday. “Ah, at last somebody that sees some of the light” easily comes to mind!

As a Suit myself, I just had to move to Linux a year ago (following the motto of “cannot understand unless you get your hands dirty”). Most surprising result (beside being able to compile kernels, understand the kids and run my own Apache server) was the lead to the Open Source Movement and all the thinking people crowding the newsgroups, Slashdot, Linux Journal and Cluetrain—easily the biggest and most interesting social experiment ever!

Being an inhabitant of the old world, about 14 hours away from Silicon Valley, the Net is a boon. Not being able to keep mum I moved a few of my own scribblings (more to come..) onto a web page for the heck of it. If you ever have the time it can be found on http://www.thingamy.net/. The last one, “The Shortest Info Age Business Guide Ever” reflects “my humble view” on some of the issues that you covered in your article in the February Linux Journal. BTW, a recent experience with the power of the Internet was when I hacked out a short piece on “A Suit's Experience With Linux” for a local Linux site on a Sunday evening (Feb 6th). Somebody linked it to Slashdot and by the time I got up next morning I had 30 personal mails in my box and in excess of 400 comments to the article on Slashdot. Kind of a revelation for somebody used to old fashioned business...loved the feedback!

Thanks again for sharing and best regards,

—Sigurd Rinde, sigurd@rinde.com

Moving to SMP

Michael Keller wrote a very good article concerning Multi-processor OS's, but I'm afraid it's quite apparent the editors let at least one typo through. In his first paragraph, he writes: “Windows 9x users need not bother, since that platform supports only one CPU per host.” I'm sure he must have meant not ”per host“ but “at most”.

—Greg Leach , gleach@lingerie.com

[LJ March 2000] Canon Fodder comment

In an otherwise good & interesting article on brand naming, there's one minor but troubling statement; Doc Searls states in the second paragraph that “...Red Hat Package Manager (written by Caldera, now used by pretty much everybody)...” To the best of my recollection (and I was involved with both distributions at the time) RPM was developed in house by Red Hat - period. It grew out of the old RPP system, and was incorporated into the Caldera beta releases and Caldera Open Desktop system, the first commercial release from Caldera.

Caldera _DID_, at least in the first phase when the distribution was based on & derived from Red Hat 2.0, help fund some of the RPM develop- ment, and Caldera personnel probably have made contributions to RPM, it being an open source project. But RPM was not, to the best of my knowledge, written by Caldera.

—Rick Forrister, rickf@crow.jpl.nasa.gov

I take full responsibility for this one. I misunderstood a statement made by a Caldera employee. I added the remark to Doc's article in an effort to be fair. However, it was a misjudgement to do so. The issue is not quite black and white: Caldera did help with the project, but perhaps only through funding as you suggest. Marc Ewing desires all the credit he gets for the development of RPM. I apologize.
—Marjorie Richardson, Editor in Chief

Best of Tech

I noticed a less-than-good answer in Best of Tech Support section of the march issue of LJ. Tim Allison (tallison@logicon.com) asked how to “log out” of linux, and talked about using shutdown(8). Marc replied with an answer that will leave Tim sitting at a linux login prompt, not getting back to LILO. The answer you should have printed was: switch to a text console, and press ctrl+alt+delete. Wait for the machine to reboot. (That's my interpretation of what the guy wanted.) The other way to accomplish this is to have a set-uid program which runs shutdown -r now, such as the following:

// I wrote this a while ago, but I don't use it anymore.  (why would I want
// to reboot? :)  This code is in the public domain.
#include <stdio.h>
#include <unistd.h>

int main(){
   // printf("UID == %d\n", getuid() );
   // printf("EUID == %d\n", geteuid() );
   // printf("UID = %d\n", getuid() );
   execl("/sbin/shutdown", "shutdown", "-rf", "now", NULL);
   perror("Uh oh, can't exec /sbin/shutdown!");
   return 1;
You can make this executable only by a certain group if you want to restrict access! OTOH, he obviously trusts the machine's users to be non-malicious, if not clueful :)

Hopefully, one of you will pass some wisdom along to Tim about the three-finger-salute method of shutdown.

Also in BOTS this March, Marc mentions that SMP is supported in 2.2.x kernels, and implies that it wasn't supported before 2.2.x. This is irrelevant, since anyone who doesn't have linux yet will be getting a 2.2 kernel at this point, but it is not true. 2.0.x supported SMP. (2.2 made in an option in the *config stuff, 2.0 you had to edit the Makefile.)

I've been meaning to mail LJ about Marc's answers for a while, since he often seems to dash off an answer without thinking about what is being asked. (You did well this month, Marc. :) Previous months have had worse errors and wrong answers that probably didn't help the person who asked :( I could dig out some examples, but I've got some sympathy for you since I don't have time to dig through my stack of LJs... :) I think it would be a good idea to have someone look over the answers given to make sure they are in the right direction at least. It wouldn't take long, just 15 minutes or half an hour to make sure that the answer doesn't have any silly mistakes and that it matches the question. Not to specifically dump on you, Marc, but your answers have had more than your share of bugs. I think it's great that anyone takes the time to answer questions at all, but a wrong answer can confuse people more.

Since I'm emailing you anyway, I'm wondering why you wrote March's Take Command column. I've never used shar or uuencode to send files by e-mail, since everyone uses mail user agents which know about MIME attachments. I just attach the file and mutt or pine takes care of it for me. Your suggestions are applicable to usenet postings, where attachments aren't an option. It seems to me that telling people they need to use shar archives or uuencoding to send files by e-mail might scare them away from trying to use anything other than hotmail (ugh).

Anyway, I hope my suggestions help make LJ better :)

—Peter Cordes , peter@llama.nslug.ns.ca

The alienation of the Linux consumer...

I am not to proud of the way that Linux Journal, you sister online mag - Linux Gazette, and you competitors have alienated the Linux consumer. There are many, many new product out there for the consumer to get there grubby hands on and proclaim to the world that Linux is here to stay. Yet since I placed my subscription with you all I have seen very few reviews of true consumer product, heard any of the latest breaking news of consumer needs, or truly as much about the consumer market as there should be.

It is true that you last issue brought up a Linux power MP3 Juke Box and I commend that gentleman for his ingenuity. And I have to say that I love these type of articles.

My problem is that there is no true mention of such things as graphics and sound card comparison, new game titles (Heavy Gear, Myth II, & heretic II), an intro to the new crop of Mail order (online and catalog) distributors, etc.

I do like the things that i have seen for the business end of the market but if Linux wants to grow the consumer market needs a carrot now and again. Sure Red Hat, Caldera, SuSE, and VA have stirred up a lot of action but there has to be the consumer base after the business Wagontrain Makes camp and the dust settles.

Thank you for your time.

—Matthew James, Mdjami0409@aol.com

France's later move to Linux ...

In LJ #71 (March'2000 issue), Dr Giovanni Orlando wrote : “France moved to Linux some months after the rest of Europe”. What's that ?

When I was still a student, in late 1991, i heard a teacher speaking of the various Unixes you could install onto your PC if you wanted to exercise at home, instead of staying in the classes (the university was all-SunSparc'd, with the exception of a dozen [MS-DOS] PC's and Macs for spreadsheets and those who were reluctant to LaTeX). This teacher said Minix was too minimalistic, Linux would exploit the newest 386 hardware but was still unusable and 386BSD the right choice. So the existence of Linux was known in France at its very first development stages. Some French developpers even contributed then.

In 1995, a serious IT professional magazine wrote an article about Linux and its seriousness. All followed. At the very same time (within the week), i discovered that O'Reilly books (Running Linux, Linux Network Administrator's Guide) had already been translated into French, that some websites had been active into mirroring TSX-11 and Sunsite for more than two years then, translating some HOWTOS and other LDP parts into French, and producing their own French documentation. In 1995 still, there were at least two French Slackware-based distribution that were localized (support for French keyboards and ISO-Latin1 in the kernel and any tools, including GNU- and X-Emacs). The editor of one of these having later been chosen by Red Hat for localizing Red Hat Linux 5.x. Finally, still in 1995, many French ISP were using linux to run their servers and routers (others used FreeBSD or commercial Unixes, one even used NeXTStep). At that time, English-or-French-written books on Linux began to blossom in France from French, Anglo-Saxon or German authors (even German books, once translated into English, were almost instantly available in specialized bookstores).

If you read older LJ issues, you can see that some French people authored articles on Linux-based application. Famous examples are those describing the use of Linux in Lectra Systems's tayloring machines, and the design of a new GUI for that purpose.

Maybe Linux trade generally did not catch early in France. There are several reasons for this. First, Linux was long perceived as a way to have Linux for free at home, for training purposes, or to avoid learning Windows for Unix-people. So few people would buy it. If they ever wanted to buy something different, techies would then choose BeOS or OS/2 Warp. Second, buying documentation that could be downloaded for free was not appealing. And third, in 1995, i could read all issues of LJ in two neighbour public libraries, and various books about Linux there, for free. The CD-ROM's coming with the books could be lended by these libraries for up to two weeks, so why should we have bought anything? Linux existed in France, but there was no way to measure that.

Another reason why Dr Orlando states France came later to Linux is that French people would not consider an Italian hi-tech company as a source for quality, and so would not buy from it, except for tight budget reasons. I know this is as stupid as a US citizen vaccinating against paludism before coming to France ; you are more likely to get paludism in Florida or New York than in Paris, where the climate is too mild. The same way, French people stick to the old stereotypes, represented by Fiat's cheap low-quality motorcars, or Olivetti's once dubious PC's, and Mercedes's expensive long-life cars or SAP's powerful software. This is stupid because Italy (which has always produced quality products along to cheap ones) also has quality motorcars (Ferrari, Maserati, ...) and well-known Linux masters, while Germany's GM branch (Opel) has become in the last decade a company that produces cars worst than Fiat's, with a quality only suited for the American market. Yet sereotypes are still there and French people would spontaneously trust any training coming from german SuSE rather than from italian FTLinuxCourse.

In the last five years, i have seen all the major american, british, german and french distributions available in France, Red Hat, SuSE and soon Corel being sold localized under their original brand names, and mountains of documentation about Linux. Maybe that didn't measure as money, but money measures nothing about using GPL'd software and docs anyway.

Only dedicated magazines appeared lately, as if the press industry had waited for our politicians' new interest in free software, a consequence of the NSA reconversion into industrial spying in the benefit of US industries leaders and at the expense of US allies industries and US small companies, of Microsoft unfair practices as revealed by the recent trials, and of the supposed placement of NSA agents inside Microsoft, suddenly making all closed-source US software highly suspect).

I thought a dottore would think enough to avoid stating something that, by essence, cannot be proven. It is impossible to know which country came sooner or later to Linux in Western Europe, because you can find contributors, users and evangelists in these countries from the very first days, and because there simply is no way to measure that.

—Arnaud alnoken@mail.dotcom.fr

Will not renew my subscription; remove me from your database

My subscription to _Linux Journal_ has come due this month. I will not be renewing my subscription.

I have subscribed to _Linux Journal_ for many years. However, I have come to realize that _Linux Journal_ is, for the most part, not helping the Free Software Movement that inspired the creation of free operating systems.

First, the magazine is quite clearly about an entire operating system, and calls this operating system “Linux” in its title and its articles. Yet, the operating system which many people call “Linux” is not really Linux. Linux is actually the kernel, one of the important components of the system. The system as a whole is basically the GNU operating system. The programmers of the GNU Project are its principal developers, but when people call the system “Linux”, the GNU Project gets very little of the credit for this. (More on this issue is available at: http://www.gnu.org/gnu/linux-and-gnu.html).

Second, Linux Journal accepts advertising and frequently promotes (through reviews and announcements) proprietary software for GNU/Linux systems. GNU/Linux thrived on the ideals of the Free Software Movement, which promotes making an ethical choice for freedom, by using only free software. Proprietary software companies are enticing us to give up that freedom and use proprietary software on top of our free operating systems. To maintain our freedom and to continue supporting the development of free software, we must resist the urge to run proprietary software on our free systems. I cannot in good conscience, subscribe to a magazine that does the opposite—promotion of proprietary software.

I encourage _Linux Journal_ to change its name to _GNU/Linux Journal_, to stop accepting advertising from proprietary software companies, and to never publish articles that are reviews or announcements for proprietary software.

I will happily resubscribe once these changes are made. For now, my personal commitment to the Free Software Movement demands that I no longer fund your magazine.

—Bradley M. Kuhn, bkuhn@ebb.org

Viewable Source != Open Source

Suddenly, it's 1984, again. That was about when the developer community figured out that access to merely _view_ source code was not enough, and they need not put up with the disadvantages of proprietary licences. Proprietary code is fine for those who like it, but might suddenly become unavailable for further improvement and adaptation if (say) the owner withdraws the product, changes business models, or goes bankrupt.

I say it feels like 1984 because of David Penn's 3-Mar-2000 article, “Tripwire Opens Up 'Best of Breed' Security Tool” (http://www2.linuxjournal.com/articles/business/034.html). Penn reports that Tripwire, Inc. will be “providing source code for its flagship product, as opposed to merely open sourcing older versions....”

However, “providing source code” _isn't_ open sourcing. Am I missing something, or isn't this free / open-source world's key, fundamental difference? Did I just dream the last sixteen years? Did, say, Sun Microsystems's SCSL suddenly become an “open source” licence, merely because it makes covered source code open for inspection?

In fact, it is clear that Tripwire, Inc. remains under the (mistaken) impression that viewable source = open source: Its FAQ (http://www.tripwire.org/faq.html) states that “Tripwire, Inc. has had the advantage of distributing an open source product in the market for 8 years”. This refers to “Tripwire ASR”, the viewable-source variant of the company's product. Which, of course, is not open source, and never was.

Examples abound, actually, of the open-source community going to considerable lengths to replace proprietary, viewable-source software, to gain the advantages of genuine open source. The canonical example would be BSD Unix and its progeny. In the security field, GNU Privacy Guard is replacing proprietary PGP, OpenSSH is replacing SSH, _and_ Rami Lehti's GPL'd AIDE package (Advanced Intrusion Detection Environment, http://www.cs.tut.fi/~rammer/aide.html) is making Tripwire obsolete. Tripwire, Inc.'s confusion about licencing is understandable—and no

doubt genuine: They're very late to the party, are considering joining, and misunderstand the ground rules. But it's a little less easy to understand how _LJ_ could repeat the company's claims so uncritically.

—Rick Moen , rick@linuxmafia.com


I wanted to thank David Bandel for mentioning tkballistic in your Focus on Software column in the Febraury 2000 Linux Journal. This is the first time I have ever seen a shooting-related program written up in many years of reading computer magazines. As both a Linux and shooting enthusiast, I've been looking for a ballistics program that runs on Linux. This is something I'll let me online shooting buddies know about.

Thanks again.

—Dave Markowitz, dsmjd@erols.com

correction to your Corel Photo-Paint story


In your C|Net article on Corel's release of Photo-Paint for Linux, (http://news.cnet.com/news/0-1003-200-1569948.html) you mentioned Gimp and Adobe as Corel's most likely competitors. This isn't exactly true. First, Gimp has no marketing or business structure. Not even a non-profit. So, although its a terrific program, it lacks the exposure that a commercial application can get. In the long run, this may hurt it.

(I actually toyed with the idea of trying to form a non-profit or even a for-profit to keep Gimp a strong product, but coming up with a business model for this type of application is difficult. Its not likely selling the OS, where service and support can bring in significant income.)

Adobe's move into Linux is limited, so far, to its PDF and word processing tools. Its not, as far as I know, doing anything about porting is graphics or layout applications (though Frame is probably considered a layout too by many). Corel's not really competing with Adobe in graphics on Linux yet.

Mediascape (www.mediascape.com) is about ready to launch its vector based ArtStream for Linux next month. This will be the first entry into the Linux layout tools market. Not long after that, Deneba (www.denebe.com) is expected to launch their Linux version of Canvas 7, a popular Mac image editing tool with vector, layout, and Web development features. These would be Corel's main competitors in the vector graphics arena. Gimp remains a competitor in the raster-based image editing front, but the lack of prepress support and and organizational structure could eventually become a problem.

—Michael J. Hammel, mjhammel@graphics-muse.org

Proprietary Linux Training

Sair Linux Training NO
Caldera Training NO

how come i can't just buy the same materials Sair or Caldera might use in their $2000/week training courses (the binders and CDs, etcetera)? proprietary information. not to be resold, redistributed or copied. that's their very own, very peculiar trademarked way of teaching an open source OS. Phththt! to get some respect in the Linux community you just need the most comprehensive, most difficult certification test not the most expensive training classes. open software. openly available training materials.

when the Free Software Foundation comes up with certification tests as good as their free compilers, i will pay them a yearly continuing education fee and pray every day the rest go belly-up.

—Troy Anderson, satanders@yahoo.com

Vatican City *is* a country (LJ April 2000, Pg 32)


Good morning!

Hate to point this out, but the Vatican City is a country.

It has it's own stamps, passports, police, army, TLD, oh, and a seat at the United Nations.

Europe (as I guess you're American) has a slew of small city states spread all around. Sometimes it is difficult for an outsider to know if they are seperate, but as the man says, if you pay taxes to them and not their larger next door neighbours, you know they're seperate

Examples include, but are not limited to:- Liechtenstein (next door to Germany)
Monaco (south of France)
Andora (between France and Spain)
Isle of Man (between England and Ireland)
Channel Islands (actually a set of countries - between England and France)

Hopefully a small errata can get into the next issue for this!

—Tom Bourke, bourke_tm@mailhub3.davork.com

Letter in LJ

Sent to:scottbomb@hotmail.com


I just read your letter in the latest issue of Linux Journal and I'm wondering if you've investigated any Linux Users Groups in your area? I think you would be able to find some people who could show you some of the advantages (and disadvantages) to Linux. I started a group in my area two years ago and I am surprised at the range of expertise (not only in Linux) within our group. Most of us work with multiple operating systems on a daily basis and can provide some “real world” comparisons. If you get a chance, You should check into a group in your area. Check out: http://www.redhat.com/apps/community/index.html#lug

Good luck!

—Kim Henderson , kim@aerofil.com

Linux Journal April 2000.

I find myself reading with great disgust your reply to the (possibly trollish) letter published in the April 2000 edition of Linux Journal. In his letter, Scott Moore writes:

It's touted as a “free” OS (in other words, you can get it without paying a cent) but it's obviously not.

He continues to discuss his inability to find it anywhere for download, and the best reply you can muster is to point him to a download site. Frankly, I'm ashamed. Not only to you pander to his delusion but you further propogate a myth that's circulating more and more through inaccurate press and word of mouth. Repeat after me:

Free as in speech, not free as in beer.
Free as in speech, not free as in beer.
Free as in speech, not free as in beer.

Please, please, for the sake of your readers and the people currently looking into Linux, set this record straight in your magazine - free software is about FREEDOM, not PRICE. The fact that the price happens to be zero or near-zero is a nicety and most definitely not the whole enchilada.

Perhaps you felt a more thorough reply was not required due to the obvious rant-factor (perhaps my letter will rate similarly on that scale, in the opposite direction?), but this is (IMNSHO) one of the worst mistruths circulating in the minds of the general public.

—Dave Baker , dave@dsb3.com

Linux and IBM PowerPCs

As a regular reader of the Linux Journal and a happy IBM RS/6000 owner running Linux, I was quite satisfied with the March issue, and its Linux and IBM PowerPC article. I have a few comments on it, though.

First: There are availble more than one Linux Distribution in the world that will run on your IBM RS/6000, allthough Yellow Dog is the only one that is supported by IBM. The LinuxPCC distribution has a lot of the core Linux/PPC developpers in the “stable”, and has maybe the best unofficial support on Linux/PPC via its mailing lists. SuSE's upcoming release 6.3 for PowerPC will run on pretty much the same workstation models as Yellow Dog. Same goes for Debian and its coming 2.2 distro. I have successfully installed Yellow Dog, LinuxPPC, the SuSE 6.3 beta, and Debian Potato on my 7248 system without too much hazzle. There are availble step-by-step installation instructions for several of the distributions mentioned.

Second: The main reason that this is possible is of course that the Linux kernel is the same in all the distributions, so that if one distro can run on a special model, there won't be too long till all of the others can do the same.

Third: IDE based IBM workstations and laptops have some trouble running with the stable release of Linux. Working sets on the mailing lists use patched kernels from the unstable 2.3 series. Hopefully we'll have beautiful support for this machines when the stable 2.4 series arrives. Until then it's quite possible to get your IDE machine running, but prepare for some hours of work.

Fourth: The main reason that older IBM PRePs are not officially supported by Yellow Dog is quite possibly the lack of a real working X server to this kind of hardware. Because of this, the PRePs won't reach it to the desktop unless running as a server, connected to a good X terminal. Hopefully, we'll have working X servers in the future, supporting this kind of hardware.

Last: A good list of working hardware for Linux on PPC, and pointers to installation instructions for several kind of machines can be found at http://www.linuxppc.org/hardware The LinuxPPC distribution can be found at http://www.linuxppc.org , and is commercially sold from http://www.linuxppc.com Unofficial installation support can be found at http://lists.linuxppc.org (the url in the article has a typo) The SuSE 6.3/PPC beta can be found at http://www.suse.com/ppc Debian Potato/PPC information can be found at http://www.debian.org/~porter and in the unstable tree on a debian ftp mirror

Regards, Ingvar (Who is writing this on his GNU/Linux-running 7248-133 while listening to mp3 on it's built in sound adapter, running GNOME, and sending this mail via it's built in ehternet adapter.)

—Ingvar Hagelund; UiO, ingvar@unik.no

Article on NetMax in Issue #72


A comment from the trenches on your review of NetMax WebServer.

I had a less-than-stellar experience with the NetMax Professional version of this product in an [granted] chaotic environment of 11 Macs (of various ages), 2 NT workstations and 2 W95 boxes. Though perhaps adequate for stable and non-disruptive environments, it proved untennable in this one. Small configuration and hardware problems proved enormously difficult to track down, and we are reverting to a vanilla Linux distribution where config files are under our own control. Where things are behaving again.

Users should be aware that the CGI scripts that control all the service daemons so nicely presented in the web interface are completely closed-source. Moreover, many of the familiar /etc/rc.d config files are no longer relevant. Consequently anyone who relies on being able to solve network configuration problems by hand-editing config files may be frustrated. As I was.

Though the tech support responsiveness from NetMax was great, ultimately the fundamental problems of a closed system began to feel too much like my battles with M$.

A few simple examples:

—David Updegraff , dave@toimi.com

Desktop “wars”

First, nice work. I've been a subscriber for a while and I'll continue. That's a vote of support where it counts.

The main reason I'm writing is to send a note about the Gnome/KDE conundrum [including Phil's editorial in support of KDE]. Some of us don't want the windows desktop. If I wanted a “task bar” and a start menu, I'll just boot into NT [actually, regardless of whether I want them or not, I get them]. These two are not the only game in town. The series on “Artist's desktops” is a step in the right direction, but there's more to X than E or an M$ windows wannabe.

I've run 9wm, wmx, and am currently using ctwm. I like ctwm because it gives me the virtual desktops I need, sound feedback on some events [via rplayd], and allows me to configure keyboard/mouse click shortcuts as I see fit. Of course, I need to read and think and take the time to change things [an ongoing process], but I like it this way. I've always thought that twm had the right design philosophy- and ctwm fits right in there.

Now you may think I'm a hopeless UNIX geek- but even outside of my personal preferences, it's pretty easy to see how one-size-fits-all doesn't work in sneakers, and it doesn't work in desktop metaphors either.

I'm not interested in telling people what to run on their desktops, but to me, Gnome/KDE is a step backwards [toward Redmond] instead of forwards [towards individual customization].

—John Saylor , jsaylor@mediaone.net

Ranting and Raving Back

“Do you wanna know what really pisses me off about Linux?” These are the words spoken by Scott Moore in the April issue of Linux Journal. For a moment I thought that the letter must have been written by Bill Gates himself until I continued on and realized that the reason he is so upset is that he is having a hard time downloading a copy.

Is it just me or is this the absolute most rediculous thing you have ever heard? I wonder if Scott even really looked for a place to download the software or if he just touted the MS Party line and concluded that all might Bill has to say is true.

As for the lack of software available for Linux, I suggest that Mr. Moore take a look at freshmeat.net or any of the dozen other sites that I can think of where you can get source code and pre-packaged software for all the major distributions of Linux. My belief is that Mr. Moore is afraid of learning something new and is unwilling to go take more than one step beyond his install shield installer and his start button. Kudos to all of you who contribute to the production of this fine zine and to all the great software developers and systems programmers out there who have made the cult called Linux available for the rest of us who are not afraid to try something different.

—Andrew Armstrong , slipknot@highertech.net

DSL Access

I read with curious interest Jason Schumaker's article in Issue #72 of Linux Journal. I think Jason is wrong on several points.

One, you don't need a router for DSL. I directly connected my computer to the DSL modem and it's worked fine. You can hook up a hub to the DSL modem and drive more computers, but you have to get more addresses from the ISP. If you want to go the cheap route and need a router, Linksys makes one with 5 Ethernet ports, works with a DSL modem, and has a 4-port hub for $250. I've seen this router as low as $150 to $180 depending where you go on the internet.

Two, I think what Jason was lamenting about was the situation in his community. I live in Houston Texas, and I've had DSL for about a year (April 1, 1999 was my install date). Southwestern Bell services this area and I'm very happy for the service. SWBell is now offering DSL service with free installation and hardware for a one-year commitment. And you get to choose your internet provider. I ordered my DSL service through my provider, and if I have any problems, I call either SWBell if there's a problem with the line, or my provider if there's another problem. There's no questions on whom to call if the line goes down.

Three, Covad, Northpoint, and Rhythms buy the lines from the local ILEC and sell the service bundled with the DSL line. Many providers, like Internet America, use Northpoint to provide DSL services. Remember these companies are targeting businesses, not regular home users. SWBell targets both residential and business users, so I think a person would be better off going with DSL from their local ILEC.

I do agree with Jason about Cable Modems. I think the cable modems are a losing proposition because the providers assume Windows stations will be the clients. With the customers in hubs, there will be congestion just like the dial-up modems. The DSL line offers a dedicated pipe. The phone companies are eager to make more money from these DSL lines.

I think right now the decision to make is not based on which service, but rather what area one is in. Down here, DSL is dominating cable modems due to the slow rollout of cable modem service. SWBell had a 6-month head start on their high-speed service ahead of Time-Warner, for example.

Keep up the good work.

—James A. Buckner, jbuckner@texas.net

Well, yes, Jason was writing about his area specifically, but the same problems happen in other areas too. As a matter of fact, I have friends who live in Houston (my home town) who have never been able to get DSL service and have resorted to cable modem. So perhaps they like Jason, live just outside the limit for being able to use SW Bell's service. Also, the DSL modem is not a true modem, it is a type of router. —Editor

Issue 72 - “that's not an uptime....This is an _uptime_.”

68% of statistics are made up or wrong. :)

In LJ Index for April 2000, it's stated that the longest uptime for a Linux system was 498 days. That's just not right. Ask around and you'll find people who can tell you about their 500+ day uptimes. Mine ended somewhere roughly around 570 days when the system needed to be physically moved. I can't find in any of the popular search engines my final uptime post when that system was shut down, but I know I did post about it.

Here are two from shortly before that.


I've seen stories from others about multi-year uptimes with Linux. Claiming 498 days as the record is selling Linux way short.

—Jon Lewis, jlewis@lewis.org

At the forge—April 2000

Just got through reading the April 2000 issue....it was great as usual. In particular, the “At the Forge” column by Reuven Lerner was a good beginners article on SQL database design. I have one quibble, however (there is *always* an however, isn't there?): He creates a table (StationLines) that has a column north_to_south, which “is an integer value that counts the number of stops between the beginning of the line and the named station.” Doesn't this make it a pain if, as he asked rhetorically on the first page of the article, “a new station is built between Haifa and Tel Aviv?” You would have to go into the tables and renumber all of the stations that came after the new station. Wouldn't it be better design to use a constant value, in this case maybe milepost number? (Station A is at milepost 0, Station B is at milepost 10.2, Station C is at milepost 45.9; the new Station D is at milepost 15.6...we don't have to renumber anyone, and the SQL query can still pull them out in north to south order. Just a thought...

—John Quentin Heywood, heywood@wcl.american.edu

up Front - April 2000

your source for point 14 in LJ Index (498 days of the longest uptime on Linux systems) is grossly misinformed. In my own experience on one consulting jobs I had a Linux machine which at the moment we are parting our ways with this customer had already around 650 days of uptime. If they not tore down a building which is housing it (there was such possibility) it may well still be running now. I do not know. 650 days was in my understanding slightly amusing but nothing that remarkable for a server, as opposed to a “hack-box”, with UPS.

In recent discussions on linux-kernel list somebody mentioned his machine which is closing now to 1300 days of uptime. 498 is not even cutting close. It is true that 'uptime' utility counter will wrap around after roughly 498 days but this does not mean that a box stopped. The same way as a car odometer which wrapped around the third time does not make this car brand new. :-)

Michal Jaegermann, michal@harddata.com

—Michal Jaegermann , michal@ellpspace.math.ualberta.ca

Please post in the Letter to the Editor on Linux Journal

May I propose that you post this in the “Letters” section of Linux Journal. Thank you for an EXCELLENT magazine.

May I define two concepts? The “open-model” is the concept that ideas from one person/organization may freely pass to another person/organization in order to collaborate (http://www.opensource.org/osd.html). The “closed-model” is the concept that ideas from one person/organization are restricted to those within their circle of control.

History shows solid, ample evidence that both models create competition while the open-model feeds progress and the closed-model stifles progress. I hope not to detract but solidify my position, by providing a few examples. These examples can extend in every facet of life (Government, Education, Society, Art, Economy, etc.). May I detail the value of open-model in the technology facet.

In 1820, Danish physicist Christian Oersted demonstrated electromagnetism by pushing a compass under a live electric wire. This caused its needle to turn from pointing north, as if acted on by a larger magnet. Oersted discovered that an electric current creates a magnetic field. In 1821, Michael Faraday reversed Oersted's experiment. He got a weak current to flow in a wire revolving around a permanent magnet. In other words, a magnetic field caused or induced an electric current to flow in a nearby wire In 1830, Professor Joseph Henry transmitted the first practical electrical signal. Showing that electromagnetism could do more than create current or pick up heavy weights—it could communicate.
In 1837, Samuel Morse invented the first workable telegraph. In 1861, Johann Phillip Reis completed the first non-working telephone. Tantalizingly close to reproducing speech, Reis's instrument conveyed certain sounds, poorly, but no more than that. Alexander Graham Bell, on the other hand, saw telephony as the driving force in his early life and become know as the father of the telephone.

Source: http://www.privateline.com/TelephoneHistory/History1.htm

All of these followed the open-model of collaborating information and NOT keeping it closed.

I believe the closed-model is not bad; however, the open-model is a much better way.

The natural, basic, fundamental, or primal outcome of open-source is progress. Open-source breeds healthy competition, which breeds incremental improvements. In contrast with the closed-model, the open-model does not blindly reinvent the wheel but insightfully works with others and builds upon or corrects what others have already done.

Often I see arguments about which model is better. The open-model advocates have nothing to worry about so why cause a fuss? Are some open-model advocates criticizing closed-source software companies because they are concerned the closed-model will get ahead, become better, blackmail open-source, or prohibit open-source?

Close-source software may very well continue to get ahead of where it currently sits. That is fine. Let them go at their pace and open-source will exceed their pace.

Can closed-source becomes better software? Well, sure. That is their goal and some may argue they have done that. They will with us on their tails—pushing their movement—providing healthy competition. Yet
closed-source companies will compete with some of the best and smartest programmers this world has to offer.

Will closed-source blackmail open-source advocates? They have and I believe they will continue. So what! Truth, over time, will triumph. Perjury will lead to eventual death. We tend to find out the correct details sooner or latter.

Will closed-source companies prohibit open-source? If they are successful, then we have damned ourselves. “Dam: a barrier built across a flow of technological information impeding progress.” This is where we need to legally and prudently fight. The strategy to prohibit or otherwise choke the open-model is a direct attack on the core of the open-model—FREEDOM. This is where the real danger lies. We should be
concerned with this threat.

I believe the open-source advocate's focus should not be to waste time, resources, and efforts on criticizing closed-source or the closed-model principle. Where will that get us? If angry people just don't get the open-model concept, then getting them more upset won't help them. Some criticizing has lost (in the argument) the fundamental hallmark of open-source: FREEDOM to collaborate.

The entire fundamental concept of open-source or the open-model is FREEDOM to see, decide, and act.

The open-model leads to rapid success and a change of attitude for some people.

—Valden Longhurst , valden@longhurst.com

Installing SUSE 6.3 on my Toshiba Laptop

The installation was fairly straight forward.

I first changed the BIOS BOOT Priority to FDD > CD-ROM > HDD to enable the system to boot directly from the CD. This then started the installation process which I found to be very easy to use, although I was expecting at some stage to be presented with a list of available packages, missing dependencies and a host of error messages to tell me it didn't what hardware I was using. To my surprise(and joy!) with only a few easy decisions to be made, I found the installation progressing extremely well.

Once I got the system booted, I found that it would get to the login prompt and then Freeze. After some thought (and some swearing and a bit of Jack Daniels) I hit on the idea of the PCMCIA settings(don't ask me why or how) this is when I changed the BIOS settings from Auto-Selected to CardBus/16-bit and hey presto! a working, booting system.

Then came X.

I had initial problems with the Xserver config and had to fish around on the web for some assistance. http://www.cs.utexas.edu/users/kharker/linux-laptop http://www.buzzard.org.uk/toshiba http://sdb.suse.de/sdb/en/html/grimmer_cyber9385.html These three sites gave me a few pointers but I had to resort to using an external monitor in order to get SaX running when I got fed up with hacking XF86Config files manually(I could never get the syncs correct). I would definitely recommend using an external monitor for setting up X on a laptop(especially a Toshiba because they have the 'Hot Switch' FnF5)

After X was up and running, I then realised that the DHCP client was not connecting to my NT server and picking up a valid address. I didn't want t spend too much time with this so I gave the machine a permanent IP address.

I then set up Samba to use the Printer on the NT server using yast. (Easier done than said!)

I then set up Netscape and StarOffice to access my Proxy server and I now have a portable Linux Workstation.

Once this was sorted, I set about installing Sybase 11.9.2. This told me that I needed certain versions of libraries and kernels which were, in fact, older than the installed versions so I installed the packages with the -nodep option. This worked a treat!

I shall soon be getting DHCP working (although this is not a priority) and also I am going to try out my portable HP CD Writer. This should keep me busy.

I have included my XF86Config file and machine configuration below, just in case anyone is interested.


—Sean Hackett, seanh@teknosys-corp.co.uk


# SaX autogenerated XF86Config file
# This file was generated from the SaX
# Version: 2.8 - sax@suse.de
# Date: Fri Feb 25 20:56:43 GMT 2000
# Xserver:SVGA
# MouseVendor:Unknown
# MouseName:Unknown
# RamDac:207
# Dac8:207
# Dac16:207
# Dac24:
# Dac32:

Section Files
   RgbPath  /usr/X11R6/lib/X11/rgb
   FontPath /usr/X11R6/lib/X11/fonts/75dpi:unscaled
   FontPath /usr/X11R6/lib/X11/fonts/100dpi:unscaled
   FontPath /usr/X11R6/lib/X11/fonts/Type1
   FontPath /usr/X11R6/lib/X11/fonts/URW
   FontPath /usr/X11R6/lib/X11/fonts/Speedo
   FontPath /usr/X11R6/lib/X11/fonts/misc
   FontPath /usr/X11R6/lib/X11/fonts/75dpi
   FontPath /usr/X11R6/lib/X11/fonts/100dpi
Section ServerFlags
Section Module

# This section is no longer supported
# See a template below
# Section XInput
# EndSection

Section Keyboard
   Protocol Standard
   XkbRules xfree86
   XkbModel pc104
   XkbLayout   gb
Section Pointer
   Protocol    PS/2
   Device      /dev/psaux
   SampleRate     60
   Emulate3Timeout   200
   BaudRate    1200
Section Monitor
   Identifier  Primary-Monitor
   VendorName  !!! LCD !!!
   ModelName   XGA 1024X768@60HZ
   HorizSync   31.5-50
   VertRefresh 58-62
Modeline 1600x1000 104.00 1600 1616 1968 2080 1000 1000 1007 1044
Modeline 1280x960 83.20 1280 1296 1552 1664 960 960 967 1003
Modeline 1024x768 66.00 1024 1040 1216 1328 768 768 775 802
Modeline 640x480 25.79 640 656 720 832 480 480 484 501
Modeline 1600x1200 104.00 1600 1616 1968 2080 1200 1200 1207 1253
Modeline 1280x1024 83.20 1280 1296 1552 1664 1024 1024 1031 1070
Modeline 1152x864 74.80 1152 1168 1384 1496 864 864 871 902
Modeline 800x600 40.35 800 816 928 1040 600 600 606 626
Section Device
   Identifier  Primary-Card
   BoardName   TRIDENT
Section Screen
   Driver   SVGA
   Device   Primary-Card
   Monitor  Primary-Monitor
   DefaultColorDepth 16
SubSection Display
     Depth  32
     Modes  640x480
SubSection Display
     Depth  24
     Modes  640x480
SubSection Display
     Depth  16
     Modes  1024x768
     Virtual   1024 768
SubSection Display
     Depth  8
     Modes  640x480

Section Screen
   Driver   Accel
   Device   Primary-Card
   Monitor  Primary-Monitor
   DefaultColorDepth 16
SubSection Display
     Depth  32
     Modes  640x480
SubSection Display
     Depth  24
     Modes  640x480
SubSection Display
     Depth  16
     Modes  1024x768
     Virtual   1024 768
SubSection Display
     Depth  8
     Modes  640x480

Machine config.

Manufacturer : Toshiba
Model : Satellite 4060XCDT
Hard Disk : 4.3 Gb
Memory : 131072KB
Display Chipset : Trident Cyber 9525 2.5Mb
Network Card : 3COM PCMCIA CardBus 3CCFE575BT
   Domain Controller is an NT SBS machine
   Currently also have a Red Hat 6 server running a development Sybase 11.0.3

BIOS : ACPI version 7.60
Bios Settings
First Page

Power On Display  = Auto-Selected
LCD Display Stretch  = Disabled

Battery Save Mode = Full Power

Pointing Devices  = Auto-Selected
Ext Keyboard Fn      = Disabled
USB Legacy Emulation = Disabled
Parallel Port Mode   = ECP
Hard Disk Mode    = Enhanced IDE(Normal)

Power-up Mode     = Boot
CPU Cache      = Enabled
Level 2 Cache     = Enabled
Auto Power On     = Disabled
Alarm volume      = High
System Beep    = Enabled

Next Page

Device Config.    = All Devices

Controller Mode      = CardBus/16-bit

Serial         = COM1(3F8H/IRQ4)
Built-in Modem    = COM2(2F8H/IRQ3)
Parallel       = LPT1(378H/IRQ7/CH3)

HDD         = Primary IDE(1F0H/IRQ14)
CD-ROM         = Secondary IDE(170H/IRQ15)

VGA Segment Address  = C000H

Floppy Disk    = (3F2H/IRQ6/CH2)

PCI BUS        = IRQ11
LJ Archive