More Letters to the Editor,

Smudgy Ink and Page Borders

I got the August issue yesterday, and the letter regarding the black borders on the Feature pages caught my attention. Having worked here and there in the publishing business, I feel very qualified to offer the following alternatives: Instead of using a solid black border, why not use a 25% border with the screen angled at 60%? Less ink, and it still makes the pages stand out when looking at the edge of the magazine. Another possibility is to go ahead and keep a solid black border, but only along the top of the pages. That way, when the reader is reading it, there isn't the ink on the left and right edges where the magazine is held. Still visible from the “outside”, just not all the way around.

Keep up the excellent work. (I know you will anyway, since Linux users tend to be the type not to put up with a bunch of PR bull.)



In the August 1999 issue, #64, one Lisa Zuckerman ( wrote in asking why KDE was failing to find “libstdc++2.9”, as well as asking why vi was telling her that .xinitrc was not a regular file. The answer given by Mr. Marc Merlin ( was rather incorrect, and not very helpful. So, if I may, here is the /correct/ answer...

  1. Libstdc++ This is the standard C++ library commonly used in Linux, often it is supplied by GCC. Version 2.9, however, is supplied by egcs 1.1.X, which is not included with Slackware3.6, but IS included with Slackware4.0
  2. contains all the necessary files, and a “” script, needed to build and install KDE on a Slackware system.
  3. Switching distributions is not really necessary.
  4. About .xinitrc
First, check the permissions on the file
   ls -col ~/.xinitrc

Here is mine:
   -rwx------   1 jeremy        552 Nov  5 15:40 .Xclients

you should see something similar, most important is the “rwx” at the beginning, that means that the owner can “r”ead, “w”rite, and e“x”ecute the file.

Second, if the permissions are correct, try this

   file ~/.xinitrc

Which should produce results similar to the following:
   /home/jeremy/.xinitrc: ASCII text
If it doesn't, then it has surely become corrupted. If this is the case, back it up, then create a new one (This example assumes you wish to start KDE by default)
   mv ~/.xinitrc ~/.old.xinitrc
   echo "exec startkde" > ~/.xinitrc

—Jeremy Crabtree,

Thinkpad 750CS Article

I was glad see the article in the March issue with information on the 750CS. I bought a Thinkpad 750CS for my wife last week, but had a great amount of trouble with the floppy. I ended up buying another laptop (a 755CSE) that had a 1.44MB floppy in it and put the hard drives in the 755 during my Linux install to get at a 1.44MB floppy. Thanks for the info on the floppy. I thought I had a useless 2.88MB disk drive until now.

I've since put a Raytheon Wireless LAN card in both systems to make mobile web terminals. Very nice, I can go to the pool and surf the web or reach the other laptop if it is nearby.

I have a few question though. One is _when_ to run tpdualscan... I can get it to work if I run tpdualscan manually then startx or xdm, but when I try to put it in my /etc/X11/xinit/xserverrc or rc.boot it runs but X looks the same as before.

Also, any luck with the on board sound? I identified the chip set as an mwave, but have had no luck in getting it to work.

BTW, I got the 750CS for $99 and 755CSE for $298 at a local refurbishing shop. I love these old laptops, and they work great under Linux... not too many platform can run on a 486 and perform so well. ;) Again, thanks for the great article, and keep up the good work.

Chris McClimans,

Opinion: LJ and the future

Hi, I'm a new LJ reader. I like your magazine but I have the following comments (and I hope, some helpful tips):

Now I realize this probably isn't news to you and that you are caught between keeping existing readers happy with the current format and trying to gain a bigger market with a different pitch. However, the Linux phenomenon is growing all the time and I believe more and more consumers will want a magazine with the aforementioned characteristics. The challenge for The Linux Journal is to inevitably evolve along with the Open Source Revolution. I feel that it is perfectly positioned to be THE AUTHORITY on Linux product reviews ala PC Magazine in the Windows world.

In closing, I hope that my comments are helpful even if you don't necessarily agree with them. Here's hoping LJ becomes a big success!

—S.McPherson, smcpherson@microworks.a

“Letters to the editor”

After visiting the Red Hat, SuSE and Debian web sites, I realized that everyone seems to avoid posting any graphic icons of “tux” the Linux penguin.

Its interesting how commercialization of Linux can cause people to avoid using the primary Linux logo (tux).

An interesting article for LJ might be “Linux Marketing Strategies & Tactics” or something similar...

I think if we stop and take a look at how Linux is being marketed, we would see disturbing trends that are emerging everyday..

Larry Ewing, the creator of the tux logo has a web site:

—Shiloh Costa,

LTE letter

George Saich
—*Cheers*+ for your well-considered essay in the LJ e-mail column! While not out of place, your apology for your “extreme” stance earns my verbal support.

It is my belief that Extremism in the Persuit of Computing Freedom is no vice. I would suggest that something like California's Community College System should be recruited to provide this training (trainers need State Instructor's Certification, and students gain a modest State subsidy) at minimal end-user expense (and the “required manuals” for courses should be LDP-standard downloadable documentation) in keeping with the Free Software Spirit.

Do other states have Community College, continuing-education-type, public-access education systems set up? Here in the SF Bay Area, anyone can take a good, 5-unit college-credit training course for one quarter for well under a hundred dollars. Similar training courses (taught by private training groups) cost _many_ hundreds and often cram the whole session into three weeks, usually requiring students to take time off work. The difference between these two approaches is bloody obvious!

—Eric Palmquist,

Article Review for issue 2!

Article: Linux System Administration: You as System Administrator, Issue 2

I just read your 3-pager on System Admin' on the net. I sincerely enjoyed your style of writing.

As an educator in this industry for over 15 years ( CNI/MCT, others) I pride myself in the ability to “never have loose scope” of the first-time users perception of a particular subject matter,... especially one of this nature and magnitude. The analogies used to “link” the concepts of the Unix/Linux file system using the DOS environment were excellent. Thank you for publishing this on the net!

I am currently evaluating other such documentation as well as co-developing as series of courses in Linux ( re-inventing the wheel to “my” way) and look forward to reading more of your work...Again..Thanx!

—Tom Foster,

Article: “The Point Really is Free Beer”

I am astounded that your usually top-class publication has included an editorial such as that named above in your July issue of Linux Journal.

To my reading it is extremely inflammatory, making sweeping general statements that lack an appropriate level of reasoned or logical exposition (which I would normally expect of your magazine). It appears to be (in my opinion) one person's petty ranting about his feelings on the GNU General Public License (GPL), shrouded in flamboyant “high moral” talk of public good.

I guess the title really sums it up - to Mr Hughes, the most important thing is “no cost software” as opposed to software guaranteeing the (long term) freedoms and rights of an individual to be a moral citizen (as exemplified by the GPL license).

Of course, it is up to Mr Hughes to show his true colors (which I believe may well be bright) when his company's “soon to be announced” first products and services are released. If he “has been worried about software infrastructure ever since” I suspect he may well offer something wholesome to society. I am however sceptical, based on the apparent contradictions (or perhaps just misunderstandings) in his editorial.

The proof will be in the 'beer' that he offers. If it is just 'beer', ie. software provided at “no cost”, yet restricting my freedoms as a moral, caring member of the human race, I personally will not touch it (nor will any of the 25-odd employees under my management at my 'development company' (to use the same petty legitimizer)). However if Mr Hughes' ultimate intentions are positive and good, I will be the first to support his work. Certainly I have heard only positive information about the Cypherpunks, of which he was a founder. We could perhaps all take heed not to trip over our own exalted egos (yes, myself included).

Please put your “editorial advisory board” (as listed on page 4) to work in ensuring the top quality content usually expected of your magazine.

Sincerely (and a little (in)“flamed”, and surprised),
—Zenaan Harkness,


I am rather disturbed by several articles that appear in the May 1999 issue. If Linux Journal continues to publish programming articles by technicians who work “at a local computer store” or book reviews by high school students, I won't be renewing my subscription again. The first case is “Introduction to Multi-Threaded Programming” by Brian Masney. A quick look at the sample code provided is all it takes to see what kind of “authority” Mr. Masney is. Besides the uninitialized pointer in main() that produces an immediate segmentation fault it is quite clear that this guy never learned a thing about neatness. Publishing this kind of code will never do anything to promote readable code. As for “Review: Linux Programmer's Reference” by Richard Petersen I am completely dumbfounded as to why reviews are being handled by high school students. I am not trying to denigrate Mr. Petersen's knowledge or ability but I expect articles in this magazine to be written by authors who are more experienced and can provide a viewpoint that is more relevant to those of us who have spent years in the trenches on major league projects. If LJ can't come up with contributions by more credible authors I don't think that I will be the only one deserting. Just my 2 cents.

—Bill Lewis,

Beginners section

With all the new people looking into using Linux, I recommend that Linux Journal add a larger section for the novice Linux user. If you keep this section separate from he rest of the magazine, the experienced Linux users can easily ignore it. However, it would be very beneficial for new Linux users.

—joe lerch,

Issues with the August issue...

First, my compliments on a nice magazine. I've been a subscriber since the second issue. I have always appreciated the clean approach you have taken with the publication.

That said, I'm a bit disappointed at the direction LJ seems to be taking. The first thing I noticed was the foldout advertisement for Penguin Computing. ARGH! I thought LJ would remain pure. I detest the standard computer rags for the use of reply cards and “fat ads”. The first thing I do if I ever read one of those magazines is rip out all the cards and ads. I suppose this is an economically driven decision to keep subscription prices down, but please don't do it. I would rather pay an extra couple of dollars per year to avoid this annoying, wasteful, scourge :)

It seems the layout is moving to that of those “other” publications as well. Your “Up Front” section is difficult to follow. Trying to forge a path through a myriad of sidebars, arrows, boxes, etc., makes it difficult to pick out the important information. This practice doesn't seem to have infested the main part of the journal (the articles) and for that I'm grateful. Please limit the font-o-philia and layout fluff to the the front few pages. It felt like I was watching MTV with frequent cuts, fast pans, and no content. Ouch! (A minor beef: the lengths of the bars on the bar graphs depicted do not correlate with the numbers indicated. Confusing.)

Also, I second a reader that noted the smudging of feature articles with the thick black border. Perhaps some of us excrete solvents through our thumbs that are effective in dissolving the ink used in LJ?

My last comment is on the “Best of Technical Support”. I'm rather dismayed at the level of the responses. Many of the questions chosen for printing are fundamental newbie questions, but often the responses are cryptic, cursory, and probably do more harm than good. Not that I purport to be an expert, but these responses should be screened for accuracy and completeness before being printed.

Ok, enough griping. I applaud your work and will continue to read LJ. Just please remain true to the clean, informative approach. I'm sure you never hear from the silent majority that has no pet peeves worth complaining about!

—Steve Singleton,

A Letter for the “Letters” Column

I must respectfully disagree with Troy Davidson in his letter, “Standards”, of the August 99 issue.

As to the HP 722c printer, I dare say it can be made to work with any Linux distribution. I use an HP 672c at home with Red Hat 5.1, and these 2 printers are fairly similar. Try, and I wager that within 2 days you will have info on exactly how to fix that printer. This web forum is tech support beyond anything commercially available.

I don't personally see the Linux community as trying to beat or destroy Microsoft (at least not overtly). Linux appears to me to be a collective effort to produce the best operating system and associated applications possible. The motivation behind this effort is more of an artistic challenge than an economic target of opportunity.

If Linux is better than NT and Win98/5, and I believe it is, then it's only a matter of time until it wins over the market share on its own merits—without a declared war against Microsoft.


—Bill W. Cunningham,

Linux Expo

Hello, my name is Dave and on the 24th of may I sent the message below to the Linux Expo coordinators. I still have not received a reply and I am still disgusted at the unfair treatment I received at the expo. Since the expo, I have stopped using Linux, stopped promoting Linux, and switched to OpenBSD. I feel Linux as a whole is turning into a huge moneymaking scheme just because it happens to be the popular “flavor of the day”. I will continue my subscription to Linux Journal, in my opinion, it is simply the best computer magazine in publication. Thank you for your time.

Sent to on 24 May 99—Hello, I attended the Linux Expo this weekend with my two daughters ages eight and seven. You can imagine my shock when I was asked to pay full admission price for them. How can you expect to charge for people who don't understand anything that's going on inside? I paid the price, because I wanted to attend but was unable to purchase anything as laying out the additional $40 for my two daughters cleaned me out. I was very upset by this and DO NOT plan to attend next years event. Unless, of course, I am given a refund of the $40 and you change your policy of charging full admission for children.

—David J. Pote SSgt USAF,

Article in August Linux Journal

I was disappointed to see that your article “Graphical Toolkits for Linux Programs” overlooked the very powerful and free Fast Light Tool Kit (FLTK). FLTK is an excellent GUI widget set with corresponding GUI builder that works natively with both the win32 and the X/11 API. It also extensively supports OpenGL. The widgets are very high quality, fast and compact. The whole library is released under the LGPL, which makes it a much better choice for commercial development than Qt or Motif.

Check out for more information on this excellent, free GUI widget set and GUI builder. Sincerely,

—Darren Humphrey,

Multi-Threaded Programming

I have a comment regarding the good introductory article on page 74 of your May 1999 issue.

Mr. Masney states “...threads...will automatically take advantage of machines with multiple processors.” Based on my experience, this is not [yet] true.

I used pthreads to design & code an image resampling benchmark. I ran this program on a 14-processor SGI (IRIX64 V6.2) & on a 12-processor Alpha (OSF1 V4.0). In both cases vendor-OS specific function calls had to be added to ensure each thread executed on a different processor. The SunSoft document Multithreaded Programming Guide indicates the same is true for Solaris 2.4.

In conversation with systems analysts at the facilities I used, I learned that the POSIX pthread design did not include standards for implementing multi-processing. I do not know if it has since been updated.

I initially developed the benchmark algorithm & program on my 486/Linux desktop (no threads). I ported the code to both SunOS (no threads) & Solaris (pthreads) inside of an hour. The code quickly ported to the SGI & Alpha requiring minimal consultation with system analysts. The best performance was achieved using vendor thread libraries. Conversion from pthreads to vendor threads required no major code alteration.

—Robert Geer,

PCI Modem letters-to-editor

In response to Greg Bailey's correction concerning SOME PCI modems being Linux-friendly, I feel compelled to point out that the the problem lies in “WinModems” - modems designed to operate only with MS Windows, through a LICENSED INTERFACE to MS Windows. These modems are often bundled with systems or sold very inexpensively. The fact that 3Com has a “generic” 56K PCI modem with a street price of $100 and a PCI “LoseModem” for only $60 makes me wonder what incentives may exist in the marketplace for making Windows-only modems.

Thanks for the great magazine!

—Peter Cavender,

PS/2 mouse problems FYI

I have installed them both, and have found a common problem in both. I think it is a kernel problem....

I have a Compaq 5630 Presario with 1 serial com1 port, 1 ps/2 mouse port, a parallel printer port, with plenty of memory and a quantum bigfoot noisy as hell ultra IDE drive.

I also have an OEM Compaq branded 2 button ps/2 mouse , and a 3 button MouseSystem dual-mode serial/ps/2 “white” mouse both mice work perfect in “all” modes of operation both serial and ps/2 on my windows partition. Same can not be said of my Linux partitions.

I began by installing Red Hat with my MouseSystem mouse on the serial port. This was a good and lucky choice, as I was later to discover. After installation it occurred to me that I might need to use my serial port for something else, so I tried switching my MouseSystem mouse to the ps/2 port and changing my config to use it. Dead meat city, as soon as I tried to use the mouse on the ps/2 port, the entire keyboard locked up and my system was dead. This was Red Hat 6.0.

Later when I tried to install caldera with the fancy lizard install which tries to detect the mouse, I had the same problem.

This occurs also with my Compaq branded 2 button ps/2 mouse.

There are a ton of problem reports in the caldera user-forum archives about problems with ps/2 mice.

One person made the observation that the “paraport” driver was reporting the detection of a “ps/2” device... This would seem like a big red flag to me shouting KERNEL PROBLEM !!!

I don't understand why caldera does not pass these problems along to somebody who works with the kernel or its drivers, so that it can get fixed.

Could you please notify somebody in Kernel land, that we are having a problem out here in the wilderness

—Steve Bovy,

“PCI modems”

I sure thought by now someone would have pointed out the problem, but from the “Compounding Errors” letter and response, it doesn't look like it.

All you said about PCI modems would have been right, if only you had used the term “Winmodem” instead. I don't doubt that some (or maybe even all, I never saw one myself) Winmodem actually uses the PCI bus, but that's rather irrelevant to the problem.

A Winmodem is a modem that doesn't include most of the CPU and software needed to do anything decent; instead, it has software (typically only for Windows) to get the host CPU to do it instead. This supposedly makes for a cheaper modem. Unfortunately, you pay for it by having a chunk taken out of your main CPU, but hey, we all know Windows users don't *really* multitask anyway, right? And nobody needs docs on how to do this for other OSes, because Windows is all there is, right?

One guy recently claimed on linux-kernel to have access to the necessary docs and source to write a driver for one specific Winmodem for Linux. I have no idea if anything will ever come out of this, however.

Oh, and there also seem to be Winprinters around with the exact same problem, I hear.

So: please don't blame the PCI bus for this!

—Kai Henningsen,