Letters to the Editor

vi, most used text editor

First of all, let me tell you how much I appreciate LJ, for the accuracy of the information it gives and the nice tone it uses.

I write you in reaction to the ``1997 Readers' Choice Award'' that the text editor vi received in the December issue. I'm an old addict of this program, and I still find it fast and useful. But I have always promised myself that I would learn Emacs one of these days, thinking I couldn't remain an old dinosaur using such an old tool. The problem is that I always found Emacs far too complicated. So when I saw the Award, I was genuinely surprised so many people think like me and I had a good laugh.

In short, after reading LJ, I decided that, despite vi being an old and ugly editor, I'll keep using it without any remorse for being an Emacs loser.
--
Pierre-Philippe Coupard
coupard@mipnet.fr

Fujitsu Lifebook

I read the review of the Fujitsu Lifebook 420D in Linux Journal Issue 43 (November). Maybe some of the LJ readers would be interested in this information.

Since March 1997, I have owned a 520D, which uses a Chips&Technologies 6555x for its video chip set. It runs XFree86 just fine with up to 64K colors at 800x600 (virtual up to 1600x1200@8bpp), and acceleration using the SVGA server. I don't know if this model is still current; however, it's almost the same as the 420D except it does have an external floppy adapter.
--
Jeroen Beekhuis
j.beekhuis@uci.kun.nl

Mixing Linux and NT

``Best of Technical Support'' in Issue 44 (December) suggested a way to mix Linux and NT. Here's another way using the NT boot loader. Assume NT is on /dev/hda1 and Linux is on /dev/hda2, and that /dev/hda1 is mounted on /dosc.

1. In Linux, edit the /etc/lilo.conf file so that:

   boot=/dev/hda2
   

2. Run the following:

   /sbin/lilo
   dd if=/dev/hda2 of=bootsect.lin bs=512 \
   	count=1
   mv bootsect.lin /dosc
   

3. In NT, edit the file c:\boot.ini to include (make it the default if you wish):

   c:\bootsect.lin='Linux'
   

Jumbled Text and Setting Color

I have a few comments in regards to ``Best of Technical Support'', November 1997. Shells like bash and zsh perform their own line editing, and to do this they need to know how long the prompt is. With Jim's PS1, bash thinks the prompt is 15 characters longer than it actually is, because the colour setting escape sequences don't take up any screen space. This can easily be seen when typing a command which reaches column 65, because bash assumes it is actually at column 80 and wraps the text at that point. The real solution is listed in bash's man page under PROMPTING:

\[ begin a sequence of non-printing characters, which could be used to embed a terminal control sequence into the prompt.
\] end a sequence of non-printing characters.

\[\033[36m\]\u_\[\033[33m\]\W_\$_--\[\033[32m\]_

Yes, you are correct. My solution does address one problem but upon re-reading Jim's question, I realize that it will not solve his. Your solution is the correct one. Thanks Carey.
--
Chad Robinson
chadr@brt.com

Linux in the Mainstream?

I was reading the article in Linux Journal #43 called ``Linux in the Mainstream?'' by Phil Hughes. As a Linux user I do like the idea of Linux becoming more mainstream. However, I don't agree with the suggestions that Mr. Hughes makes. Linux is something different, that's why we use it--well, why I and many others use it. Linux started as a small project and was taken on by a large community. It is not like other operating systems. There is no need to make Linux mainstream--Linux is becoming mainstream on its own power. This is not because of marketing--it is because Linux is a damn fine OS, and people are finding that out by word of mouth. We, as a community, should not look into marketing and certification programs, but rather we should keep right on with what we have been doing--telling others about this great OS. We don't need a central authority to hand out meaningless sheets of paper.

I am not saying that Linux is not lacking in some areas. It most certainly is, but as a community we should be working on the shortcomings simply for the sake of making a better, free, OS--not to win over new users.
--
Steve Carpenter
sjc@delphi.com

SCO Unix Review

I really appreciated your recent review of SCO OpenServer. Where I work, we recently had reason to take a computer that had Linux on it and make it into a dual Linux/SCO system. I would like to point out, however, an error in the review. Ken says:

While you're going through this process, OpenServer is merrily overwriting your master boot record and wiping it free of LILO.

While it is true that SCO overwrites LILO if you have it installed on the Master Boot Record (MBR), it is not true that LILO cannot boot SCO. In fact LILO is more than happy to boot SCO. The problem is that SCO expects its own partition to be active or bootable. From the README file for LILO:

Some PC UNIX systems (SCO and Unixware have been reported to exhibit this problem) depend on their partition being active. Such a setup can currently only be obtained by installing LILO as the MBR and making the respective partition active.

If, after you install SCO, you reinstall LILO to the MBR and make the SCO partition bootable, LILO will very easily allow you to choose one or the other at boot time. On our setup, we have Linux installed to /dev/hda2 and SCO on /dev/hda4. Our lilo.conf file, therefore, looks like this:

boot=/dev/hda
map=/boot/map
install=/boot/boot.b
message=/boot/boot.msg
prompt
timeout=100
#
# Linux partition
image=/boot/vmlinuz
        label=linux
        root=/dev/hda2
        read-only
#
# SCO Unix partition
other=/dev/hda4
        label=sco
        table=/dev/hda

Perl Option Error

Thanks for publishing my message in the ``Letters to the Editor'' in the December 1997, Issue 44. But you introduced a huge mistake in it, which can have security implications for readers who blindly trust LJ.

The message, published under the title ``Big Brother'', mentions the -T option of the Perl interpreter, saying that ``-T tests that the file type is text, not binary.'' This is ridiculous and I never wrote that. I wrote that every Perl CGI programmer should use the -T option and explained that it refers to tainted mode (man perlsec for details). The -T option (a command-line flag) has nothing to do with the -T function (which indeed tests if a file is text). Any Perl programmer could have caught that mistake.

It seems to me that the treatment of my alert message (remember that anyone on the Internet could execute any command on a machine which uses the scripts you originally published) exhibited two serious flaws:

It was treated too slowly. Most people trust paper more than Usenet News or WWW. Many people probably assumed that the articles in LJ were carefully scrutinized and that the scripts were dependable. LJ had, in my opinion, a responsibility to warn users as soon as possible (at least in the next issue) of the mistake and not through a letter to the editor two issues later.

It is perfectly understandable that you edited my message; I know that my English is quite poor. But you could have sent it back to me for a last check. I do not think it is ethical to modify a message, not on a grammatical point but on a technical one, and to publish it without showing to the readers the edited parts and without sending it to the author for proofreading.
--
Stephane Bortzmeyer
bortzmeyer@pasteur.fr

First, let me apologize for your letter getting changed in a way that changed technical content. We try hard not to let this happen. One of our copy editors thought the -T needed more explanation and obviously grabbed the information from the wrong place. I agree he should not have added to the text without consulting you. If you had put as much detail in the first letter as you did above, I don't think he would have felt he needed to add anything. Ultimately, though, I did let his addition pass, and I take full responsibility for the error.

LTE is just about the last column I put together. Consequently, there is not a lot of time to pass it back and forth. It is also the first time I even see the letters, so they can be old. By the time a magazine comes out, the next issue is already at the printer, so errors never get corrected until two issues later. It's too bad, but such is the way of magazine deadlines.

Actually, I think you do quite well with your English.
--Editor

Fan Mail

Just wanted to let you know that Linux Journal continues to be startlingly good. I'd say that 70% of your articles are of immediate interest to me each month, and the other 30% get re-discovered as useful knowledge when I go through my back issues. It's like you guys can read my mind! Keep up the amazing work.
--
Manni Wood
mwood@sig.bsh.com

From the Publisher

I got a chuckle out of your ``From The Publisher'' article on page 10 of Linux Journal, Issue 43 (Nov 97). It sounds as if IBM is relegated to the past. We have three IBM AS/400s (64 bit RISC) and 600 users, mostly on Windows 95 clients. IBM has passed Intel and Microsoft in both hardware and software terms (stop by http://www.as400.ibm.com/). OS/400 (V3R2 and later) has everything a company needs to do relational database Intranet/Internet applications. I'm using Net.Data (which is available on many platforms; see http://www.as400.ibm.com/netdata/) to do Provider lookups, map engine address lookups, etc. The inclusion of Java in V4R2 and IBM's business class libraries makes the AS/400 a very reliable database platform for the future. We have one table with over 28 million records, and I'm not quite ready to trust the MS SQL server to handle it.

How do you get Linux to be more mainstream? You have hackers like me who are established in the corporate world. I'm installing Linux to test JBuilder Java applets at home. (I want applications to run on AS/400, NT and Linux without modification.) I have a small network with NT server 4, NT workstation 4, Windows 98 beta and a P150 class laptop with NT workstation. I'm dumping my Windows 98 machine (486/66) to load Slackware Linux. I'm hoping for much better performance.
--
Steven P. Goldsmith
fcci@cirs.com

SAMBA

I was disappointed with the article ``Using SAMBA to Mount Windows95'' in the November issue. It gets a number of basic facts wrong.

For a start, SAMBA does not do what the article claims it does. The author is actually talking about a kernel-based SMB file system called smbfs written and maintained by Volker Lendecke (among others). Volker is a member of the SAMBA Team, but smbfs is definitely not a part of SAMBA. SAMBA is an SMB file server portable to all Unices, whereas smbfs is currently for Linux only.

The article also says that ``SAMBA is a program that allows Linux to talk to computers running Windows for Workgroups, Windows 95, Windows NT, Mac OS and Novell Netware.'' The bit about the various Microsoft operating systems is partly right (although he actually meant smbfs) but the part about Mac OS and Novell Netware is totally wrong. For those he is probably referring to MARS_NWE, ncpfs and Netatalk, which are totally separate packages that talk totally different protocols. Perhaps if you are publishing articles on a topic which Linux Journal editors are not very familiar with (there is a lot to know about Linux), you should send a copy to someone who is familiar with the topic so they can do a quick check and point out any obvious errors.
--
Andrew Tridgell
andrew.tridgell@anu.edu.au

Well, you caught me--I don't know everything about Linux. However, to make up for this, I do send articles to copy editors who have given me their areas of expertise. The copy editor who worked on this article was a networking expert. I'll keep you in mind for the next SAMBA article that comes in. --Editor