"Linux Gazette...making Linux just a little more fun!"

Contents:

• Running Multiple Instances of X
• VC Madness
• Linux and OSPF
• Security Problems with pop3
• Cryptographic System
• An Interesting De-Referencing Problem
• Reminder!
• pcmcia ide Drives
• KDE BETA 1
• Compression Program
• loadlin
• WipeOut

Running Multiple Instances of X on One Video/Monitor (VCs)

From: Guillermo S. Romero [email protected]

Hello,
I have tried to run multiple X servers with only one card and one monitor. Is this possible, or is it normal that the second X server does not run? I used startx display :0 the first time, and :1 the second. I have a 1024K video board (#9GXE64 PCI, S3 864), and normal config is 8 bpp, 1024*768 virtual desktop, running on a remix of RedHat 4.0, 4.1 and 4.2, with XFree86 as server. Maybe I did not understand the man page (English is not my first languaje). Any suggestion?

The normal way this is done is using the form:

	startx -- :0 &
		startx -- :1 &

... The -- is used by startx and xinit to separate an optional set of client parameters from the set of display/server options and parameters.

If you ran the command:

	startx xterm -e myprog -- :1 &

... it would start X Windows with a copy of xterm which would be running 'myprog' (whatever that might be). The remainder of the line informs the X server to use display number one (which would be VC -- virtual console -- number eight on most Linux systems).

(On my systems it would start on VC#14 -- accessed with the {Right Alt}+{F2} key combination. I routinely configure mine with 24 VC's -- the first twelve of which have "getty's" (login prompts) and the next eleven of which are available for X (xdm's or otherwise), using 'open' commands, or for dumping status output from a process (like 'make' or 'tail -f').

Read the man pages for startx and xinit one more time. I'm pretty sure that the man pages have all been translated into Spanish -- so you might want to hunt those down.

Thanks!!!

Read the man pages for startx and xinit one more time.

Sure, and with a dictonary. ;]

I'm pretty sure that the man pages have all been translated into Spanish -- so you might want to hunt those down.

Try:

man-pages-es-0.2-1.src.rpm: ftp://ftp.redhat.com/pub/contrib/SRPMS/man-pages-es-0.2-1.src.rpm

The Spanish Howto is small, too global, sure it does not cover that. And I still have problems with my ntilde chars and acents, Spanish is not supported a lot (Linux or another OS, always late and bad)... The system explained in that howto does not work (but thats another question, whose solution maybe... magic? real support?). GSR

I'm afraid I'm completely ignorant of internationalization issues with Linux. I do know that there is quite a bit of work done on Linux boxes in Japan, Germany, Italy and, naturally enough, Finland (where Linus comes from).

As bad as it seems -- Linux' support for other languages is probably the best in the world. Unfortunately I don't have the skill or resources to point you to the support and resources you need.

Since your English is clearly adequate to discuss these issues with me -- you might consider contributing some of your time to a translation effort (get the LIGS, NAG, and SAG portions of the Linux Documentation project translated, and "beef up" (improve) the Spanish-HOWTO.

I highly recommend that you find or start a Linux user's group in your area. This is the best way to help yourself and to improve the situation for all of your compatriots.

-- Jim

VC Madness

From: [email protected]

Hi
I have an application that uses its own .cshrc and .bashrc to fire up and this is done by using its own login account. Now what I would really like is for this to say select VC8 to run on and then have my normal X on VC7 as usual. Can this be done? and if so how?
--Phil

	open -c 8 -- su - $PSUEDOUSER

... where PSUEDOUSER is the psuedo users whose .*shrc you want to run. Naturally you can convert the .*shrc into a normal shell script and do whatever you like with it. You have to run this as root -- (so 'su' doesn't prompt for a password) though there are ways to get around that 'runas' is available at the sunsite.unc.edu archive site and its mirrors). If launch this from another UID you'll need to ensure that this users (the launching user, not necessarily the psuedo-user) has write access to /dev/tty8 (group +w should be sufficient).

If you want to have the console visually switch to this application's VC you can just add the -s switch like so:

	open -c 8 -s -- ....

... where the "--" marks the end of 'open's' arguments so that the command that follows it can unambigously get its own arguments.

Without the -c switch the 'open' command will select the next available VC. Any subsequent 'startx' commands or other 'open' commands would then pick later ones (unless the others were freed back up).

You can have two or more copies of X running on different VC's as well. For example the command:

	startx -- :1

... will create a second X session on the localhost:1 display (the first one is addressed as localhost:0 or simply :0). These X sessions can be run under different UID's and have completely different client configurations (colors, window managers, etc). There is also an 'Xnest' command that works similarly -- allowing one X session to run "within" (as through a window on) one of your existing X sessions.

You can also set the terminal settings and colors using normal redirection of the form:

stty erase ^? > /dev/tty8

... and:

setterm -background blue -foreground yellow -bold on -store \
		> /dev/tty8

This last command would set and store a new set of default screen colors for the VC. The setterm command can also be used to control the Linux VC screen blanker's timeout (a value of 0 means "never blank").

Naturally you may want to read the man pages for all of these.

If you want to ensure that a given process will *always* be running (and will automatically be respawned when it dies) you can add it to your /etc/inittab -- so that the init process will watch over it. This is how new 'getty' processes are spawned on your first six (or so) VC's when you boot and are respawned when you logout. Likewise if you use 'xdm' to keep a graphical (X based) login prompt on one or more of your VC's.

As you can see, its possible to do quite a bit with Linux VC's. I run 12 VC's with getty (as login consoles), have one 'xdm', one devoted to syslog, and ten more available for other purposes (such as 'startx' and 'open' commands and to to use for 'tail -f' commands when need to monitor the end of a status or log file -- from a 'make' or whatever.

The second set of 12 VC's is accessed with the *right* {Alt} key. (In case you'd never noticed, the default keyboard settings of Linux only allow you to use the *left* {Alt} key for switching VC's). I set syslog to use VC number 24 with an entry in the /etc/syslog.conf file that reads:

	*.*			/dev/tty24

This puts a copy of *every* syslog message on to that VC -- which is what I switch to for a quick glance and try to switch to when I leave any of my systems unattended. (That way when one does lock -- as rare as that is -- I have some idea of what the last throes of the system were).

I set that to bright red on black with the following command in my rc.local file:

setterm -foreground red -bold on -store > /dev/tty24

(I also do the same to /dev/tty12 which I customarily use only for root login's).

Hope all of that helps.

-- Jim

Linux and OSPF

From: Jose Manuel Cordova-Villanueva [email protected]

Dear Sr.
Recenty I had my first contact with the Linux G. and is a big source of information, can you inform me if there are a program that can talk ospf because our ISP, is changing from RIP to OSPF and we have a linux box in one of our links, for our cisco no problem but for our Linux box??

The software you want is called 'gated' (for "gateway daemon"). This is a Unix multi-protocol router package for Linux which includes support for OSPF and other routing protocols (BGP4, IGRP, etc).

Here's a link to the top level 'gated' pages Cornell Gated Consortium Information

I've heard that compiling 'gated' for Linux is not quite trivial so here is some other links that might help: Here's a link to a source RPM in the Red Hat contrib directory: gated-R3_6Alpha_2-1.src.rpm

Here's a threaded archive of the 'gated' users mailing list: Gated-People Archive Here's an odd note about an alternative routing software package/project: Route Servers -- RA.net: routing arbiter project

Hope that helps.

-- Jim

Security Problems with pop3 of Linux 2.1.29

From: Sam Hillman [email protected]

Well I hope I'm posting to the right person. I have two questions, which I hope you can answer. 1. How do I setup my linux machine as a POP3 server? I can't find any FAQs or Howtos.

Usually you don't have to do anything extra to allow access to POP services. Most Linux distributions include a pop server pre-installed and appearing in the /etc/inetd.conf and /etc/services files.

A quick test is to login to the system in question and type the command:

telnet localhost pop-3

+OK your.hostname .... (some copyright info)

If that doesn't work you'll want to make sure that the appropriate lines appear in your /etc/services and /etc/inetd.conf files like so:

/etc/services:
pop-3		110/tcp	# PostOffice V.3
pop		110/tcp	# PostOffice V.3

/etc/inetd.conf:
pop-3   stream  tcp     nowait  root    /usr/sbin/tcpd	ipop3d

If they appear commented out -- remove the leading hash sign(s) (or paste these samples in) and restart your inetd with a command like:

kill -HUP `cat /var/run/inetd.pid`

2. When I log on to my ISP, I download my mail and it gets dumped to the sendmail, this creates a situation where the mail is bounced back and forth until it passes the hop limit and is dumped as an error message in the postmaster box, and a nasty letter is send to the originator from MAILER-DEMON... I think this maybe because I'm running a local area network between my two machines, the IP address of the local net is 162.blah.blah... But I also have the IP address the ISP gave me in the host file. If the ISP's IP address is the problem can I remove it from the host file, and just get a duynamic IP when I connect? Thanks in advance!

This is a bigger problem. First the 162.*.*.* is probably not what you want to use for you disconnected LAN. There is an RFC 1918 (originally RFC 1597) which describes and reserves a set of addresses for "non-Internet" use. These are guaranteed not to collide with any valid (routable) hosts on the 'net. Here's the list of those addresses:

	192.168.*.*  (255 class C address blocks)
		172.16.*.* through 172.31.*.* (15 class B address blocks)
		10.*.*.*  (one class A address block)

The next problem is configuring sendmail for use on a disconnected system. You probably need to define your hostname (or an alias to your hostname) to match what your ISP has named you. Each ISP seems to use a different way to manage these "disconnected sendmail subdomains" -- with no standardization in site (which is why I use UUCP). I gather that some people use a scheme where they only run sendmail when they are connected. The rest of the time their MUA (mail user agents like elm, pine, mh-e, exmh, etc) just drop outgoing mail into the mqueue directory where 'sendmail' will get to it later. One problem I have with these configurations is that sendmail wants to look up these remote hosts. This seems to cause various problems for users of "disconnected" or "periodically connected" (dial-up) systems. So far the only solutions I've found are: recompile sendmail without DNS support (there used to be a sendmail.cf switch that disabled DNS and reverse DNS activity in sendmail -- but that doesn't seem to work any more) use UUCP. UUCP was designed for disconnected (dial-up) and polling systems. It's what I use. The disadvantage to UUCP is that it's a bit hard to set up the first time -- and you have to find a provider that's willing to be your MX/SMTP to UUCP gateway. There are still some people out there where will do this for free or at only a nominal fee. But they are increasingly hard to find. I use a2i Communications in San Jose. You could use a non-local provider if you want to use UUCP over TCP as the transport mechanism (UUCP is pretty flexible about the underlying transports -- you could probably use tin cans and string as far as its concerned).

There are several HOWTO's that try to cover this topic. Try browsing through some of these:

• ISP Hookup HOWTO: Basic introduction to hooking up to an ISP.
• Electronic Mail HOWTO: Information on Linux-based mail servers and clients.
• Mail Queue mini-HOWTO: How to queue remote mail and deliver local mail.
• Offline Mailing mini-HOWTO: How to set up email addresses without a dedicated Internet connection.
• UUCP HOWTO: Information on UUCP software for Linux.
• Sendmail+UUCP mini-HOWTO: How to use sendmail and UUCP together.

-- Jim

Thanks so much for the detailed suggestions. We have installed a newer version of pop3 on our server for now and we will look into the feasibility of implementing some of your suggestions for a final cure. Thanks again, James, we really appreciate it.
-Sam Hillman, Service Manager, Easyway Communications.

Cryptographic System

From:Emil Laurentiu [email protected]

Hello Jim,

Sorry for bothering you but I would apreciate a lot an answer even a short one like 'no' :) I am (desperately) searching a crypographic system for my Linux box. I am already using TCFS but I'm not very happy with it for several reasons: it is slow, I experienced some data loss, must use the login password, cannot share encypted files with other users, NFS - increses security riscs. And the people in Italy seemed to have stoped work on this project (latest version is dated february).

February doesn't seem that old.

Are you sure you're using the latest TCFS (v 2.0.1)? You can find that at: http://pegaso.globenet.it (which is a web form leading to an HTTPS page -- so use and SSL capable browser to get there).

If you find it slow than any other decent encryption is also likely to be too slow for you.

You could look at http://www.replay.com (in the Netherlands). This has the best collection of cryptography software I've seen anywhere.

The two fs level alternatives to TCFS are CFS (Matt Blaze's work, on which TCFS was based) and userfs (which support a few different user-level filesystem types including an experimental cryptographic one.

I am wondering if you know anything about an encryption at the file system level. Something like SecureDrive (from DOS :) which did IDEA encryption on the fly at sector level for a partition and was very fast.

Are you sure SecureDrive is using IDEA? I rather doubt that.

As an (almost) single user on my linux machine something like this would be more apropriate. Of course if I would not find one I'll finish by writing it by myself. My only concern is that I've been a Linux user only for half a year and I did not get the chance to study the kernel to well (this will be a good opportunity :)

Why not pick up on the TCFS or CFS work? Why not build on the userfs work (plugging in whatever encryption you like)?

Why write it "by yourself" when you can collaborate with other members of the Linux community as they have done to bring you Linux itself, and as the FSF and others have done to bring you the GNU packages which turn Linux into a full OS?

What you asking for doesn't need any support at the kernel level. userfs and CFS already have shown that. The Linux kernel already support a robust and open filesystems interface (which support more different filesystem types than any other -- with read-only support for HPFS, NTFS, BSD and Sun UFS/FFS, and support for HFS (Mac), ext2fs, xiafs, Minix, and many others.

If you're a competant programmer (which I am not, BTW) you should be able to trivially take the sources for any of the existing filesystem modules and hack together your own with the encryption support of your choice. How secure the result will be will be a matter of your skills -- and should be greatly improved by peer review (by publishing your work for all to see).

Naturally if you are in a free country you can share your work on cryptography with the world. However the USA doesn't appear to currently be free in this particular respect -- please find a congress critter to vote out of office if this oppresses you.

-- Jim

An Interesting De-Referencing Problem

From: Kevbo [email protected]

Here's the brain teaser I read about and promptly forgot the solution (because I thought it would never happen to me). How does one delete a FILE named ".." I have the following at the root directory.

• /. a dir (which is normal)
• /.. a dir (which is normal)
• /.. a file (which is not normal)

How this happened I don't know. How to remove this not-bothering-me file has me stumped. Got an answer?

I suspect that this file is actually named something like: "/.. " (note the trailing space!).

In any event you can remove this with a command like

	find / -type f -maxdepth 1 -name '..*' -print0 | xargs -0 rm
-i

Note: you must use the GNU versions of find, xargs, and rm to ensure that these features (-print0, -0, and -i) are available. (They may be available in other implmentations -- but you must check first).

The find parameters here specify files (not directories, symlinks, device nodes, sockets, or FIFO's) and force it to only search the named directory (or directories if you list more than just /). The -print0 force it to be written as a null-terminated strings (thus the receive process on the other end of the pipe must be able to properly interpret null-terminated arguments -- which is what the -0 to xargs accomplishes).

As far as I know there is no way to legally get a NUL character into a Unix filename. (Using a hex editor might get one in there -- but fsck would probably complain on its next pass).

The -i on rm is just a little extra protection to prevent any other unexpected side effects. It forces rm to interactively inquire about each argument before removing it.

-- Jim

Reminder!

From: George Read [email protected]

I am a subscriber to caldera-users, but as a rank newbie, 99% of what gets posted is irrelevant to my situation and over my head. In fact, I'm looking for some real basic, preliminary information:

Perhaps you should consider some avenue of paid support. there are a number of consultants and even a 900 support line.

Also, if you have access to IRC there are a few #Linux "channels." (If you've never heard of IRC -- or Internet Relay Chat -- then think of it as an online CB system -- similar to the "chatboards" and "chat lines" on various BBS' and online services (like CompuServe and AOL)). Granted IRC is a bear to figure out -- and 99.9% of what's written there is even less relevant or comprehensible than the traffic on this list. However the feedback is immediate and there are some people who will take time out from their usual chat aggenda to help.

There's also that pesky "Answer Guy" from Linux Gazette ;) (but he's too ornery and doesn't help with X Windows stuff at all).

1. A way to create a primary Linux partition on a drive that is entirely occupied by a dos active and a dos extended partition. The extended partition has 400MB available that does not have any data on it, but LISA 2.3 does not wish to give it a primary partition.

You have three choices here. You can repartition the drive using traditional methods (backup your data, reformat, re-install DOS and all applications, restore data). You can repartition using FIPS (a non-destructives partitioning program for DOS -- written by Linux or FreeBSD users from what I gather). If you use FIPS the process goes something like: do a backup, verify your backup, unfragment you DOS partitions, run CHKDSK and/or Norton Disk Doctor and/or SCANDISK, then run FIPS.

Another approach -- and the only one I know of that doesn't involve repartitioning -- is to use MiniLinux or DOSLinux or XDenu. These distributions (of which DOSLinux is the most recent and must up-to-date) are designed to run on a UMSDOS partition (an MSDOS partition mounted under Linux with support for some Unix semantics). You would be running COL -- but you would be running Linux.

You can find information about DOSLinux at Kent Robotti's home page: ftp://wauug.erols.com/pub/people/kent-robotti/doslinux/index.html (Kent is the creator and maintainer of DOSLinux).

2. a workaround to a problem with RAWRITE3: I can't see my COL Base cdrom on a Nakamichi MDR7 jukebox that is controlled by a BusLogic 946C, because Autoprobe can't find anything and I can't get RAWRITE3 to write MODULES.IMG to a floppy on A:.

Have you tried supplying the "max_scsi_luns=7" parameter to the kernel during the bootup sequence (at the LILO prompt).

Normal SCSI controllers support up to 7 devices. It is possible for these controllers to refer to "Logical Units" on any/all of these devices. These "logical unit numbers" or LUN's aren't very common -- but are used by CD changers (which is why most of them are limited to 6 or 7 CD's) and some tape changes (though those usually use a different mechanism to control tape changes and ejections) and some RAID subsystems and CD-ROM "towers."

I have a NEC 7 platter CD changer which requires this parameter. This suggestion assumes that the problem is isolated to the CD drive -- and that your kernel (LISA's) is seeing the BusLogic card. If the problem is that you can't even see the SCSI controller -- then you probably want to look for an alternative boot/root diskette set and boot from that.

One of the nice things about user's groups is that you can often have the phone numbers of some local Linux users that will cut you a custom kernel on request and let you pick up the floppy. I'd highly recommend finding (or starting) a local LUG. I've occasionally had people come over to my place where we could plug them onto my ethernet and suck all the free software they want across from one of my systems.

(Which reminds me -- I've been meaning to get PLIP working for a couple of years now -- I should really get around to that). For these reasons, I ask: Is there any way to ask caldera-user users for some help on these two questions, sent to my own email address, and not have to read 20 or 30 messages that I can't profit from, at least until I get COL up and running. I had hoped from the name that Post-Only might be such an address, but I see that it is something very different.

Caldera has some support options. I think some of them are extra cost items. Have you called them about your Caldera specific questions?

At first blush it doesn't look like Caldera's COL is the best Linux distribution for your needs. If you're intent on using COL -- and particularly if you have a business need for Linux -- I'd recommend going out and buying an additional drive. For a couple hundred bucks (US) you can get a 2Gig external SCSI drive (www.corpsys.com if you don't have a suitable vendor handy).

Even if you're just experimenting with Linux and don't want to "commit" to it -- an extra external SCSI drive with a couple of Gig of space is a handy investment for just about ANY operating system. It's pretty convenient to connect the extra drive, and just make a copy of everything from your main system.

If your time is worth more than $20/hr you can easily make the case for buying a $200 to $300 hard drive. Doing full system and data backups, and verifying them prior to repartitioning can be pretty time consuming. Even if you already have a scheduled backup habit (let's face it -- most don't) and even if you have a regular recovery test plan (which almost nobody bothers with -- often to their detriment!) -- doing a major system change (like repartitioning) almost requires an extra "full" backup and test cycle.

(I have customers who've run the cost vs. time numbers for their situations and justified buying a full system and hired me to do the configuration on the same basis. The "extra" system becomes part of the recovery plan for major system disasters).

-- Jim

pcmcia ide Drives

From: Alan C. Sklar [email protected]

I am trying to install a pcmcia drive through a kit I purchased.. I got the drive all ready I formatted it with a desktop machine and bot my win 95 and linux partitions are defiend... But now when I go and boot linux I send the commad ide2=0x170 and it loads it identifies the right drive but I get all sort of errors... Can you help?
C. Alan Sklar

I don't have enough information to help with this one. Is this a laptop or a desktop with a PCMCIA adapter installed? In either event what is the make/model of the system?

Do you have PCMCIA support installed and built into the kernel? What modules do you have loaded? What does your /etc/pcmcia/config.opts file look like? What type of hard drive is this (make and model)?

-- Jim

KDE BETA 1

From: Eric Wood [email protected]

This should be the most handy tip known to man! If a certain application (I don't care what it is) complains about missing a library and you know that the library it's wanting is in a certain directory THEN:

1. Add *that* directory to the /etc/ld.so.conf file.
2. Rerun /sbin/ldconfig

That's it. What is does is it tells Linux to search the directories specified in /etc/ld.so.conf for library files. Forget about the stupid LD_LIBRARY_PATH variable. Everyone: Please read the ld.so man page for further knowledge.
Eric Wood

I recently trashed my /etc/ld.so.cache file and had forgotten how to fix it (since the last time I'd had a damaged ld.so.cache was on an old Sun a couple of years ago -- and I've never had one on a Linux box before.

Post that to your tech support archives:

System hangs on boot -- even with -b and single switches -- or it gives messages like "unable to open ls.so.cache" in a seemingly endless stream:

		Run /sbin/ldconfig!

-- Jim

Compression Program

From: Cygnus [email protected]

Anyone know of any programs for linux that decompress multi-part (multi-disk) .zip archives? I can't find a one.
-Cygnus

Most Linux distributions come with the free zip/unzip package. Here's the -L (license) notice from my Red Hat 4.2

"Copyright (C) 1990-1996 Mark Adler, Richard B. Wales, Jean-loup Gailly Onno van der Linden and Kai Uwe Rommel. Type 'zip -L' for the software License.

Permission is granted to any individual or institution to use, copy, or redistribute this executable so long as it is not modified and that it is not sold for profit."

I think there's a source package for "Info-zip" also floating around. I don't know if this is Info-zip or an independent version -- looking in /usr/doc/unzip*/COPYING I find Mr. Rommel listed -- and that document is definitely about Info-zip.

For the future you might try the 'locate' command -- which is fairly common among Linux distributions. The command:

		locate zip

... will quickly find every file with "zip" in the name or path that was on your system during the last "updatedb" run (which is typically a cron job that's run nightly).

-- Jim

loadlin

From: Scott Williams[email protected]

Answer guy, To run LOADLIN I need to have a copy of the LINUX kernel on one of my dos partitions, and an initial swap space. No where can I find an actual explaination on how to do this correctly.

You don't need an initial swap space to run LOADLIN or to load the kernel. If you have 8Mb or more of RAM you don't "need" to have a swap space at all -- but you'll probably want one. You can swap to a file or a partition -- or even several of each. Assuming that you don't have Linux installed yet you can view man pages for most Linux/GNU commands, functions, packages, and configuration files at: http://www.ssc.com/linux/man.html

... in particular you want to read the mkswap(1) and the swapon(8). The man pages there are accessed via a CGI script so you have to post data to a form to access the individual pages. Thus I can't give URL's directly to the pages in question. That's an unfortunate design decision by the web master at SSC -- it would be more convenient to access (and cause less server load and latency) if they used a cron job to periodically update a tree of static HTML pages and saved the CGI just for searching them.

Every time I try to copy the kernel to a dos diskette, Linux overwrites the formatting. DOS then cannot recognize the file from the LOADLIN command.

It sounds like you're using 'dd' or RAWRITE.EXE to prepare these diskettes. That's fine for transferring boot/root images -- but has nothing to do with LOADLIN. To use LOADLIN.EXE you copy the kernel image to a plain old DOS file.

I haven't even gotten far enough to think about creating an initial swap space... Any advice on the subject?
Scott

I'd consider getting a copy of DOSLinux from ftp://ftp.waaug.erols.com/pub/people/kent-robotti/doslinux.html (Yes there are still some people out there serving HTML pages via FTP -- nothing in the HTML spec *requires* that HTTP be used as the transport mechanism).

Kent Robotti has been working on this distribution for awhile. It takes about 32Mb of space on a DOS partition -- and comes as a set of six 1.44Mb files (so if fits on a half dozen diskettes). You then add a kernel for SCSI or IDE use.

Basically DOSLinux works like this. You get all the RAR files (RAR is a Russian Archiving Program like PKZIP, SEA ARC, ARJ, LHARC, ZOO, or whatever). The first image is a self-extracting file (an archive which is linked with a DOS binary of the decompression program -- a common DOS technique among archiving programs). You put these all in a given directory and run the self-extractor (DOSLNX49.EXE as I write this -- it was at "48" a couple weeks ago) from C:\. It thenn extracts all of these images to C:\LINUX directory.

This provides a complete (though minimal) Linux distribution. It also shows how to configure a system to use LOADLIN with a UMSDOS root partition.

I realize that you may be intending on use something like Red Hat, Slackware, or Debian on a third hard drive, or a removable drive or some other device that LILO just can't see (because you BIOS can't "see" it). You can do that -- and I've done in many times (I first used LOADLIN in about 1994 for exactly that purpose -- with the magneto optical drive I still use). However, if the README's and examples that come out of the LOADLIN package aren't helping you use if for that purpose -- than installing DOSLinux may help get you rolling and serve as a vang DOSLinux may help get you rolling and serve as a valuable example. -- Jim

WipeOut

From: Falko Braeutigam [email protected]

Hi, in Linux Gazette Issue 22 there was a question about the WipeOut IDE. Your answer was that you never heard about WipeOut :-( Please check ShortBytes of Issue #19 - there is an announcement of WipeOut. WipeOut has nothing to do with xwpe. It _is_ an IDE for C++ and Java. There is just a new release -> http://www.softwarebuero.de/index-eng.html.
Regards,
Falko

This definitely counts as my biggest flub in the 10 months that I've been writing this column. I've gotten about 10 messages correcting me on this point.

-- Jim