LJ has published several articles on the use of
Linux in Hollywood and its adoption by large corporations. I would like
to see how many small businesses, like mine, run on open-source software.
I've been using Linux for the past five years and switched one of my
associates to it at the time. How many more like me are out here in
the real world? I think there are quite a few, and it would be nice to see the
Even though I agree with the Open Source movement for operating systems,
as it creates a standardized solid foundation for further software
development, what is the purpose of extending it to other genres of
software? One thing I have noticed people fail to mention about the Open
Source movement is how do the programmers get paid?
I am a university student who eventually wants
to design software for a living. What's the point of learning extremely
complex technology if somebody is willing to make it for free in his or
her spare time? It's
like asking an electrical engineer to develop the electrical system
for a public building for free because it benefits the public.
How are software engineers
going to survive out there if they're killing themselves with open
Our job is not easy; it requires a high-level understanding of
mathematics and logic and takes years to learn. Why should we do it for
Doc Searls says the software industry is growing up and becoming more like the construction business. The engineer gets paid for his work, but the building owner doesn't have to pay for a $99 client access license for everyone who walks in the building. In mature industries, customers expect to be able to hire Vendor B to fix something if Vendor A doesn't work out. If you want to make a living in software in the long run, you'll have to fulfill those expectations. You might want to read Secrets of the Wholly Grill: A Novel about Cravings, Barbecue, and Software by Lawrence G. Townsend. To get an idea of what it's about, imagine applying some one-sided restrictions from proprietary software licenses to products you might buy at the supermarket. —Ed.
Kudos on another excellent and informative issue [LJ, August 2003]. I found the “Eleven
SSH Tricks” article particularly useful.
I am writing to comment on Marco Fioretti's review of Red Hat 9,
beginning on page 90.
It noted excessive startup time for
the desktop and OpenOffice.
This is a result of Red Hat's conservatism with regards to default system
settings. There is a configuration file, /etc/sysconfig/harddrives, that
can fix this. It contains the settings to enable DMA mode,
multisector I/O, look-ahead functionality and other goodies for modern,
fast hard drives.
Enable those settings, and you will find that Red Hat 9 is quite speedy
J. Mark Brooks
In the August 2003 Letters to the Editor section, Donald Daniel states that “Pascal inspired Ada, which was awful.” Ada was inspired by a need for a good programming language, and it fulfills that need very well. I have yet to meet an Ada basher who really knows the language.
The letter titled “Fighting C++ Rumors”
actually confirms the fact that compiled C++
libraries are incompatible with one another.
To say that this problem will soon be a thing
of the past doesn't cut it; I've been waiting
20 years! For other compatibility issues,
get a copy of C++ Primer Plus by Stephen Prata
and look for the highlighted compatibility notes.
Or, stop waiting, move to Ada and don't
bash it until you know it.
If developers want to make an
educated decision as to what language should be
used, read my article in CrossTalk magazine at
Just got the August 2003 issue, the tip on piping binary data to a remote
shell was exactly what I needed to back up my somewhat over-full iBook.
Mick Bauer's LDAP articles have been fairly well written, but I have a couple of issues with Part III [LJ, September 2003]. There are, of course, the broken DNs on page 32, which are easy enough to spot if one reads the whole article; the DNs later on are fine. However, they lead into one of the most problematic situations I've seen in the directory business. I'll call this situation John Smith.
We have 24 John Smiths working for our company. One can add a middle initial, but this is just kludging it. We have five John R Smiths. We could use a middle name; still a kludge. We now have John Ronald Smith, two John R Smiths that refuse to give their middle names and two John Richard Smiths. They're not related, and in fact, are both John Richard Smith III. (This is an actual situation at a company at which I once worked, although the names have been changed.)
Using a unique identifier is considered by many to be the best practice. There's even an attribute for it in the LDAP standard, uid, which is mentioned in another context in Mick's article. This would make one's DNs look like uid=wongfh,ou=engineering,dc=wiremonkeys,dc=org. The only restrictions on uids are that they must be unique to that branch of the tree, and they really should be unique to the LDAP server. It's also a good idea to have a standard on how one makes a uid. The only bad standard I've seen is one that isn't consistently followed, although many groups prefer one never reassigns a uid that once belonged to a deleted entry.
This is a controversial topic; many people, including some of
the original LDAP server authors, are rather devoted to cn-based DNs.
However, it can cause problems that are easily avoided, for very little
cost. I've consulted for a group that uses cn-based DNs on their e-mail
server, which accepts DNs as one user name format. Several times a year,
they ask me how they can prevent e-mail intended for an ex-employee from going
to a new employee with the same name. They always get the same answer.
IMHO, it's really about how long you want to endure the pain.
Mick Bauer replies: First, I apologize for the typos you pointed out. They were caused by my copying and pasting working code from my test server, and then inconsistently replacing the test server's real domain name with my example domain name. Second, thanks very much for sharing your insights on the “John Smith” problem. You've convinced me that a uid-based naming convention is a much better-scaling approach than a cn-based system. I think the latter works fine for small organizations, but I readily admit that it can't scale as well as the uid approach. This is an excellent illustration of how complicated and subtle LDAP design and administration can be. There are no one-size-fits-all approaches to LDAP.
I was wondering what ever became of “The Toshiba
(/article/6318). It was such a
compelling story that I just gotta know if anything ever panned out
for Adam Kosmin.
Adam struck out, but Steve Oualline got a refund for the unused proprietary OS that came with his laptop. It takes persistence, but it's possible (www.linuxjournal.com/article/7040). —Ed.
I just read Marcel's article in the September 2003 Linux Journal, taking
great notice of the MythTV part. I'm going to have to toot my own horn
for just a minute here (well, and some others). With the help of the
community, and Axel Thimm, the maintainer of ATrpms and the MythTV RPMs
for Red Hat Linux, I've put together a streamlined guide for the creation
of your own MythTV system using Red Hat Linux 9.
I have the process of creating a full-blown,
fully functional MythTV system from bare metal down to two to three hours,
but then, I've done it a few times. The process is greatly streamlined
by installing nearly every component from RPMs, using apt (pvrhw.goldfish.org/tiki-page.php?pageName=rh9pvr250).
Jarod C. Wilson
In an article in the September 2003 issue of Linux Journal, actually
a Sidebar to the article “Linux Makes Wi-Fi Happen in New York
City” entitled “Pebble Linux: Debian for Wi-Fi Access Points” Kurt Starsinic
mentions the popular Linksys WPC11 wireless card. As of WPC11 v.4,
this card is no longer Linux-compatible. The Prism chipset has been
replaced with another, and Linksys technical support will not say if
or when Linux drivers will be available. They have even evidently been
told not to identify the new chipset: RealTek.
The September 2003 issue, centering on wireless networks and Linux, was quite
enjoyable. However, there was something missing that was sorely needed.
There was no mention of what wireless network cards could be used with
Linux. The Hardware HOWTO lacks any such supported list as well. After
dredging through manufacturer Web sites, such as Adaptec and Linksys,
my outlook on converting my current LAN to wireless is not a good one.
Can you list what wireless network cards were used in any of the various
There's a good card list at seattlewireless.net/?HardwareComparison. You also can check your distribution's hardware compatibility list on the Web. As the previous letter shows, it's hard to keep up with hardware in print. —Ed.
I've been a Linux Journal reader now for around seven years, and my brief
stint as a part-time copy editor was something I was proud of and look
back on fondly. LJ has been my favorite magazine since I picked up my
first issue at a local computer show at the University of Washington
in 1996 or 1997. Recently, I bought a Mac, and with it came an offer
for a free subscription to MacWorld, which I accepted. Last night,
after reading both magazines, I realized exactly why Linux Journal has
been my favorite magazine for so long. Where other magazines make me
want to buy things, Linux Journal encourages me to
try things. Your
pages are consistently filled with informative articles about projects,
not products. The September 2003 issue's focus on community networks has
strengthened my desire to build a free node, and hopefully, a community
around it. Thank you for continuing to produce an excellent magazine.
Nathan E. Sandver
Send us the location and ESSID for that community net when you have it up. —Ed.
I've subscribed over the past year and, frankly, was not going to re-up.
The reason is simply that I'm just an average guy—a home user and most
(but not all) of your articles are over my head.
But I've been following the SCO situation, and today I read the Open
Letter from SSC to SCO [www.linuxjournal.com/article/7087]. In effect, you guys are fighting for my right
to use Linux.
Even though I'm not a professional geek, the least I can do is support
people that are supporting me (and many, many others) in our ability to
use the best OS out there. Expect my renewal in the next couple of days.
I wish you all at LJ well.
I read Tad Truex's article (LJ, September 2003) “Put a Sump Pump on the Web
with Embedded Linux” and found the article interesting. I also found
myself chuckling over what the broader audience would think of such an
That said, in a bigger context, there is the field of industrial
automation and electrical engineering that uses SCADA systems for
industrial process control.
Would it be possible to feature the subject, using Linux and other OSS
solutions in a future article, using some of your experts' experience?
I just picked up the August 2003 issue of LJ, mostly because of the SE Linux article by
Russell Coker. After reading it (with much interest) I decided to surf to
the NSA site to look at some docs and possibly acquire the downloads
concerning SE Linux. Although hardly anything to do with the American
government surprises me anymore, I was slightly surprised that access
to the site is blocked, probably for non-US based IPs, as I am
accessing the site from The Netherlands; all access to the site
was blocked even the top-level nsa.gov.
Peter van der Kleut
Try Russell's site at www.coker.com.au/selinux. —Ed.
I read with interest Tom Younker's article on converting vinyl LPs to digital formats [LJ, September 2003, page 80]. Recently I have been trying to convert my wife's extensive collection of old Russian LPs to CDs. Initially, I tried using gramofile as described by Mr Younker, but a few months ago I discovered a more recent project called the Gnome Wave Cleaner (gwc.sourceforge.net). I now use this exclusively and find that it is an excellent piece of software.
I believe that the major benefits of gwc over gramofile are: 1) As Mr Younker says, “swig has progressed, but gramofile hasn't.” 2) gwc has a nice GUI interface (gwc.sourceforge.net/main.jpg), whilst gramofile is ncurses based. 3) The click and noise removal algorithms in gwc work very well and are easy to control. As an added bonus, if there is a recalcitrant click that the auto-algorithm can't get a grip on, you can zoom in using the GUI, select the offending millisecond of sound and silence it by hand. It disappears without a trace. I certainly haven't carried out extensive testing of click removal speed and effectiveness, but gwc is fast enough for me, and nearly all clicks are removed. 4) gwc has a sonogram view that is very handy for tracking down the few remaining clicks in a track.
Finally, don't just take my word for it. Read James Tappin's excellent
page on ripping 78rpm records (www.tappin.me.uk/Linux/audio.html).
He used to recommend the use of gramofile, but he now uses gwc and
describes it as “best-of-breed”.
Don, attached is a photo of my car with the California LINUX license plate installed. When I first registered for the plate around 1993, it was installed on a rather beat-up 1986 Toyota pickup. As you can see, Linux has helped us grow our business, and the plate has been transferred to a slightly nicer vehicle.
By the way, need a DSL connection from a clueful ISP?
- Machine Learning Everywhere
- Smoothwall Express
- Bash Shell Script: Building a Better March Madness Bracket
- Own Your DNS Data
- Simple Server Hardening
- From vs. to + for Microsoft and Linux
- Understanding OpenStack's Success
- The Weather Outside Is Frightful (Or Is It?)
- Ensono M.O.
- Understanding Firewalld in Multi-Zone Configurations