TABLE OF CONTENTS

1. Welcome (at last)!
2. What is Linux?
3. The UKUUG - An Introduction
4. The Linux SIG of the UKUUG
5. Alec's Linux Slackware Notes
6. YOU want to write for Linux@UK!
7. Linux Leasure Zone
8. Linux: What For? A Personal Adventure
9. We need a Logo!
10. A Linux driver for the Specialix intelligent multiport serial card
11. Linux SIG online resources
12. Linux Local Groups Guide
13. Linux Archive CDs
14. Linux Groups Around the World
15. Linux kernelitus?
16. Your Editor: Who am I and why am I here?
17. UKUUG Linux SIG Membership Application

It is with excitement and some trepidation that I am setting out this first newsletter of the Linux Special Interest Group of the UK Unix User Group.

What is Linux?

The term "Linux" has two usages, first it is the name of the operating system kernel started in 1992 by Linus Torvalds and contributed to by UNIX enthusiasts all over the world. It has also come into usage as a term for the whole software system based on a Linux kernel at its core. There are many people and organisations that have put together a Linux kernel with other programs to make a complete working set of software. Each person hold the copyright to the code that he or she has written. Linux is not public domain software. It is protected by the GNU Public Licence which (amongst other things) stipulates that the source code to Linux and all its related programs must always remain freely available but allows people to charge money for Linux, if they wish, as long as they do not attempt to limit to the redistribution of Linux. In a nut-shell this means that even if you have paid to have some Linux software shipped to you, you are free to share the software to a friend. Sharing and mutual co-operation on the development and improvement of Linux has been the driving force in producing a world class operating system that is truly open in that everything is under your control.

Linux is being used today by tens, maybe hundreds, of thousands of people all over the world. The global Internet makes it possible for people of all nationalities to freely co-operate on Linux software projects of mutual interest. It is being used for just about everything conceivable under the sun! Software development, Internet navigation, games playing, document preparation, leading edge computer research. The low cost, and enthusiastic user community of Linux makes it an ideal way to get into a real 32 bit power operating system!

The UK UNIX User Group was formed to represent users of UNIX and open systems in the United Kingdom. It uniquely caters for the needs of people within this area and, being totally funded by membership subscriptions, is completely independent of specific hardware and software vendors. All profits are used to further the activities of the organisation.

Membership is drawn from the information technology, commercial and research / academic sectors in fairly even proportions. It is representative of workers in a variety of fields including Software Engineering; Computer Manufacture; End Users; Software Houses; Universities and Research Centres.

All are UNIX users - whether they develop kernel modifications, develop applications, are involved in teaching and research, use their system in a turnkey environment or simply want to keep abreast of everything that is happening in the world of UNIX.

The importance of the UKUUG is underlined by the fact that every major UK supplier of UNIX, UNIX related and open systems is now a member of the Group. UKUUG members actively participate in the work of UK, European and American Standards Authorities including BSI, ISO and IEEE. In addition, they provide significant inputs to Standards Committees including those involved with ANSI, POSIX, SVID and X/Open.

The Linux SIG of the UKUUG

The simple answer is that the UK Unix community recognises that Linux has the right to be considered as a Unix system. The UKUUG accepts that people who choose to run a Linux based computer system share common interest and goals with other Unix users so belong within the UK Unix users group.

Accepting that many amateurs have an interest in Linux that its free and low cost nature have fostered the UKUUUG offers two a two tier system of subscriptions. Normal UKUUG corporate and individual members can choose to take this newsletter and the other services of the Linux SIG for NO EXTRA CHARGE. People who find the 65 pounds normal UKUUG membership fee beyond their budgets can now join the Linux SIG part of the UKUUG only for a reduced cost of 20 GBP per anum. Some UKUUG membership benefits including the book and CD services and attendance at events are also available to Linux SIG members. Existing members of the ACCU (The Association of C and C++ Users) can join the Linux SIG for only 15 GBP per anum.

Going back to the original question. Why not a fully independent Linux User Group? There are considerable advantages in Linux receiving the organisational support of the UKUUG. For a start at Owles Hall there is a full time administration staff to look after the administration details such as processing memberships and properly accounting for money received and spent. It is these 'dull' admin areas that volunteer and enthusiast run groups usually fall down. By going under the wing of the UKUUG so to speak the Linux SIG can start from a sound financial and administrative basis rather than having to learn how to run itself from scratch. This means that I as SIG organiser can concentrate on the business of getting a group that is truly useful to the UK Linux community and is able to make a significant contribution to Linux as a whole.

Things I wish I had known before installation

1. Be prepared to install more than once.
2. Don't install the term source package, the object package is later
3. Do install the info package
4. When you are recompiling the kernel do the things in the following order
   • i) make config
   • ii) make dep
   • iii) make zImage
   • iv) copy zImage to /vmlinuz (save old version)
   • v) run lilo
   • vi) make clean (optional)
   This is all explained in the readme but it is a long read :-(. The build directory on my system is /root/linux.

   Some newer version have make files with lilo targets that do more of this for you. Look in the top level Makefile.

   You should be sure that you have some method of rebooting the machine with an old kernel before rebooting the new kernel ( you never know!). The Slackware setup allows you to generate a boot floppy during the configuration procedure. You can also use the installation boot disk that you did the install with.

   You can run the recompile under the UNIX batch subsystem and it reduces the system impact. Follow these steps (# is the root prompt)

   #make config
   	(answer the questions)
   
   #batch
   make dep
   make zImage
   ctrl-D
   #
   

   Now you can do something else, like play sasteriods

   This all assumes that batch and at work on your system. They did on mine right out of the box

5. You need to configure ethernet, when running 'make config', for PPP and slip support.

Handy tips after installation

1. read the HOWTO's (/usr/doc/howto/*)

2. If your printer gives
                           a nice staircase effect
                                                  like this then run
   your text through the following
   
   text source | sed -e 's/$/^M/'|lpr
   

   where ^M is control-M. This adds the carriage return that some printers require. Read the Printing-HOWTO for more info.
3. When printing text from the info documentation system you can get info to do the above for you by adding the following to your .profile

   INFO_PRINT_COMMAND='sed -e /$/^M/|lpr'
   export INFO_PRINT_COMMAND
   
   NB. to enter a control-M in vi use control-v control-M
   
   

4. If you are a Demon internet customer then there a couple of tar files in /pub/linux/Demon worth getting with hints and tips.
5. To print manual pages on a printer try

   	man topic|lpr
   or
   	zless $(man -w topic)|groff -T ascii -mandoc|lpr
   

6. To read stuff that has been compressed with GNU zip (i.e. ends with .z) use the zless utility (same as zcat file|less)

General UNIX tips.

1. To find a file somewhere on the system

   find / -name 'file_name' -print
   

   In this form file_name can contain "wild cards", patterns in UNIX terms.

   Find is very useful so learn how to use it as soon as possible.

Alec Clews (alec@earth.demon.co.uk)

The way round this is for you, YES YOU, to put your hand to writing about some Linux related subject that you would like to share with your fellow Linuxers.

Rather than saying rather bleatingly "please write for us or the SIG will just disappear down the plug-hole of apathy :(" I would rather say that the self discipline of having the aim of an article to write will mean that you will get to grips better with that subject yourself. I am assuming here that you are only going to write about what is of interest and possibly use to yourself!

Linux contains areas of great power that need explanation and understanding to make use of them. If you write then you are a trail-blazer for others into regions of truly useful and stimulating software.

Judging by the size and age of some of the major packages available for Linux such as Tex and Ghostscript with their myriad utilities in the field of text processing. X Windows for graphics, the excellent GNU C compiler suite, Perl and Tcl languages and many, many others there are already thousands of people who already know, benefit form and even work on the evolution of powerful items of computer software. The marriage of the great software packages developed to the Free Software Foundation 'copyleft' principals with the Internet for free sharing and exchange of ideas and now an operating system platform in the shape of Linux that anyone can afford has opened up the future of computing. It is now a frontier territory, a free for all where anyone can play a part and make their mark for the benefit of others.

Why should I give my good ideas away? Simply because good ideas get even better by being allowed to feed on each other in a free exchange. That is the big difference between the world of Linux and the world of commercial software. In Linux people freely and openly borrow and adapt ideas used elsewhere - with due credit. This is at stark contrast with the lawyers banquet that has been going on between major hardware and software computer companies in recent years.

Gaining expertise in Linux and the sort of software that it runs will enable you to make more effective use of your computer. This is not just cost effectiveness in that there are no licence fees to pay; but being in control of your computer means that you will be able to respond better to change and solve problems with less waste.

If you have a subject that you would like to write about, if it is common, or if it is very specialised, let me know. Linux@UK aims to carry articles for all areas of Linux use from raw beginners to those who are using Linux for a single very specific task and getting the last ounce out of it to perform that task as well as possible.

However if you do want to use you Linux machine for games playing then as well as the normal 'Vs the computer' variety the natural ability to connect to the Internet and to other Linux machines make multi-player games possible and fun.

Tell us of your personal favorite games, where to get them, how to play them, what version of Linux do they need to run and to what extent to they stretch the ability of your machine and the Linux system itself.

Writing, testing and perfecting some really stonkingly good multi user games could be a great way to get into some teqniques that have very serious and practical applications too.

A multi user game could be a good proving ground for concepts such as client-server and distributed processing. What I want is some articles about what games are out there to be downloaded and played, and ones that need willing recruits to develop and play test. There is nothing wrong with using computer technology to play games. Games are part of the learning experience for children. They should also be part of our learning experience as to what a network of Linux computers and Linux enthusiasts are capable of!

Play a multi-user game over the net

When you are connected to the game you then have to choose a character that you want to be. If you choose a name that is already in use the game asks you for a password. You just have to keep trying until you find a name that has not already been used. When you manage to find one you will be asked to supply a password and then choose attributes for your character such as race (Human or Alien) and sex.

Your hard working editor has not had any time to actualy play the game :( - could somebody supply me with a fuller discription of what to do for the next linux@UK?.

Linux: What For? A Personal Adventure

Zbigniew CHAMSKI
<chamskiz@cs.man.ac.uk>

I'd like to share with you some of my experiences with Linux. Hopefully, everyone will find here something of interest, be it memories of his/her own Linux life, the relative performance of ``serious'' workstations when compared to no-name PCs, some ideas for the next extension to buy, the desire to try this Linux that everyone talks about, or simply a good laugh. Of course, all comments and pints of good ale are highly welcome.

The Past

It was in May 1992, in the final year of my PhD thesis at the university of Rennes, France, that I first heard about Linux. At that time, I was looking for a multitasking system I could install on my PC, and was already aware of several projects of free Unix clones, either immediately available ( Minix, FreeBSD) or "coming soon" (GNU HURD). As with the approaching PhD thesis deadlines the choice of the system became an urgent matter, I spent a week or two on discussions with friends who already had installed various systems on their machines, and reading all relevant newsgroups (it's good to work at a major Internet site!).

My requirements were clear: I wanted an environment I could install at home, as close as possible to the one I used at the university: a Sun-4 under SunOS 4.1.1, with X Windows, Emacs, LaTeX, XFig and plenty of GNU software (ever tried ftp-ing to ftp.irisa.fr or ftp.inria.fr?), and most important, easy transportability of source files between the Sun and the box at home. I couldn't reasonably get a Sun-4, so my 386 PC (a branded one, an HP Vectra 386 DX/25!) had to do the job, and the sizes of applications I wanted to run required a good virtual memory system. I was lucky enough to get 12 MBytes of RAM, but having heard the Sun swapping like mad with 16 MBytes of RAM on-line, I was prepared for some hard times when writing the thesis, and especially when running Emacs, XFig and LaTeX at once, all this on top of X Windows.

In two weeks the choice was made: Linux 0.95 seemed reasonably stable, although evolving at a fair rate, allowed to run most applications I wanted to run (Emacs, LaTeX, XFig, GCC) and was able to fit on my "big" 105 MBytes hard disk. I spent a few hours downloading the files, then spent several evenings and nights trying to figure out why X wouldn't run. Eventually, after reading through the embryonary HOWTO's and taking a tranquiliser (Westmalle Triple, a Belgian triple ale) I realised that /dev/console had wrong permissions...

This was in June. In October I had a fair set of applications running on my system, gained some skills in Unix system administration and still had not experienced a single system hang-up. However, 12 MBytes of swap space were a serious constraint, with some big compilations such as the JPEG library in XV taking several hours to complete with the ``hard disk busy'' LED nearly continuously on.

By that time Linux has also evolved a lot, with version 0.99 (patch level 11, known to aficionados as 0.99pl11) considered stable enough for "official" distributions to be build on top of it. Slackware 1.02 was just coming out, and I decided to use it to update my system.

With the additional software I had installed on my own I couldn't just reinstall everything from scratch, and finally decided to install packages one by one, discovering by the way what's inside Slackware. The extra space provided by an additional 80 MBytes hard disk allowed to keep the old versions of libraries and applications, and avoided spending a week on recompiling life, the entire Linux universe, and generally, everything. And indeed, the software I ported under Linux 0.95, linked statically, ran without any problems under 0.99pl11.

The amount of disk space used by the system grew a lot, but was still acceptable, and a major hardware upgrade (a borrowed 387 coprocessor) improved significantly the overall performance of my critical applications (X, LaTeX, XFig and, eeerh, XV).

The entire upgrade to 0.99pl11 took a week, including most subtle interactions between versions of applications and libraries. Immediately after the new configuration became stable, I started writing my PhD thesis. Eventually, a month later, I faced a major problem: I needed some performance numbers for a linear programming algorithm named PIP, originally implemented using 32-bit arithmetic and breaking on all realistic data samples.

The need for performance numbers was bigger than my feeling of extenuation, so I decided to rewrite the implementation of the algorithm in a bullet-proof, if not efficient way. The task was well defined: replace the original operations on 32-bit integers by arbitrary-precision integer arithmetic, debug and collect performance numbers. However, I faced two problems: I was not the author of the original implementation, and I had to choose between several arbitrary-precision arithmetic libraries. It was time to check how good Linux' debugging tools are, and how portable are the ``portable'' arbitrary-precision arithmetic libraries.

Feeling that a more robust version of PIP could be very useful to "the scientific community", i.e., most, if not all, users of the original version whom I could contact, I chose the GNU arbitrary-precision arithmetic library (GMP). I downloaded the archive from the Inria GNU repository ( ftp://ftp.inria.fr/gnu/...), untarred the whole thing and typed `` make''. No explosion, no smoke, and `` make test'' said `` All tests passed.''. No jubilation neither: at that time I already got used to the fact that GNU software was running under Linux straight out of the box.

Then I sat down in my flat, with the PC at one end of the table, the expresso machine at the other, and the Hi-Fi and two hundred CDs to feed the player just behind me. Two weeks later the arbitrary-precision implementation of the algorithm was running, and showed its first bug only half a year later, having been run out of the box on DEC Alpha, HP series 9000 and 700, and Sun SparcStations.

Two months after completing the port I successfully defended my thesis and started a software engineer job, for which the system administration and overall Unix experience gained with Linux proved more than useful. Meanwhile, I was maintaining the arbitrary-precision version of PIP, correcting several system-independent bugs and realising that the fact that the fix works on the Sun doesn't imply that it is correct ... be it because of the SunOS' allowing negative sizes for memory blocks and files.

The Present

After a one-year forced separation from my PC, I changed the hardware to a no-name 486DX2/66, got bigger disks, upgraded to Linux 1.1.59 and once again found that bug fixes tested on a Sun aren't necessarily correct. I'm still hunting a small bug in PIP, which appears only under Linux. From my former experience and the fact that the bug appears in a code running perfectly under SunOS, I'd expect the problem to be caused by a piece of code (even one line) which, although correct under SunOS, is a nonsense under POSIX. Sometimes it's good to start with more restrictive norms.

With this Linux-based experience I found rewriting someone else's software interesting and, in some sense, fun. To the point that I decided to port ``YALAA'' (``yet-another-linear-algebra-algorithm'') to arbitrary-precision arithmetic. I found once again that getting it to run correctly under SunOS is only one half of the job, the other one remaining to be done under Linux. My feeling is that the other way around it would be once again bullet-proof in the first place. But then, I was paid to do the job on the Sun...

In fact, the Linux box became the centre for most of my research work, particularly software development and publications. Given the excellent quality of the code produced when developing under Linux, and the performance of the hardware - between two and four times faster than the SparcStations at work (the University of Manchester) - I see no reason to change this situation. OK, I'm still printing articles and such at the university, but this is where I'm supposed to do it anyway!

At the time being, working under Linux 1.1.59 is a pleasure. When I decided that GNU C 2.5.8 is not enough, compiling GNU C 2.6.2, with G++ and Objective-C took maybe forty minutes, which is one third of the time it took on a Sun-4/75 under Solaris 2.3! To synchronise my ``home'' configuration with the `official'' one at work, I also recompiled GNU Emacs 19.28, only to find that the cursor is totally opaque (as in Emacs 19.22 coming with Slackware 2.1.0).

This is where I had some trouble trying to solve the problem, as the documentation of the X Windows interface in GNU Emacs 19.28 is still somewhat incomplete. I was initially planning to reward the reader who will contribute a solution to this problem with a pint of a home-brewed winter warmer, but the original offer expired: I found a LISP-based workaround. However, the new offer is ``a pint of a home-brewed winter warmer to the person who will contribute the cleanest fix to this problem.'' Send your entry to the author.

The Future

Now, what next? I'm writing this on an all- Linux PC without a single trace of DOS on its disks, listening to my CD player (not yet a CD-ROM drive under Linux - but coming soon), and dreaming of a non-Apple PowerPC system running Linux with the display subsystem matching in quality the performance of the processor... Say, a 20" monitor with a pitch of 0.25 mm able to stand the 1600x1200 resolution of the graphics card at an 80 Hz refresh rate? By the way, for some of my research applications I find 128 MBytes of swap space a bit small...

Honestly, the Sun workstation at work still has one major advantage over an average Linux system: it comes with a very good 19'' monitor using a Trinitron tube. But this too should be fixed in near future, with a 17'' Trinitron-based monitor. Oh yes, I know, it will be almost as expensive as the PC itself, but it will also be a real pleasure (and stress relief) for the eyes. Should we think of Linux as ``a pleasure in every sense''? When compared to a decent cognac (remember the Courvoisier ad around Christmas and New Year?) it only lacks some olfactory bouquet, but I think it is better this way: I don't really like any smells coming out of my computer!

Contact Address

Mr. Zbigniew Chamski, at any of the following:

• e-mail: <chamskiz@cs.man.ac.uk>
• phone: (0161) 275 6132 (daytime)
• surface mail: 48 Tower Street, Dukinfield, Cheshire SK16 5NE

Now is the chance for you artistic types to put mouse to screen and come up with a simple but eye catching logo that will effectively portray what we are about. A few points to note.

• The logo should be suitable for letterhead paper either in mono or spot colour.
• Optionally a colourful version of the logo can be produced for sweatshirts, ties etc.
• The logo should contain the phrase "Linux/UK the Linux SIG of the UK Unix Users Group"

Method of entry

The Prize

You retain copyright on your logo as long as the group is granted permission to use it on letterheads, the newsletter and other official publications both paper and electronic. Any merchandising done using the logo (it could happen) will be done with you getting royalties.

Gallery of entries

John Risby's entry (john@enville.demon.co.uk)(

Callum Potters' entries (cpotter@ipsys.co.uk)

I started work on the driver in May 1994. At the time there was no support for any Intelligent multi-port serial cards in Linux.

After looking at the existing Linux serial driver I decided to base my driver very closely on it - things like opening and closing the port are very similar. The interrupt service routine had to be redone from scratch as the Linux serial driver normally allocates a separate IRQ line for each port whereas the Specialix card shares one IRQ across all ports on a card. All low level accesses to the card are completely different and these too had to be redone from scratch.

After receiving documentation from Specialix I got the initialisation code working in an evening. Specialix also provided a fully working Unix driver as a sample but I decided not to look at it if at all possible to avoid copyright problems.

The main effort of getting the driver to read and write took perhaps another two days work. Things seemed to be progressing well when a bombshell arrived in the new kernel. The internal tty interfaces in the kernel had substantially changed! I tried to work around the changes but ended up basically starting from scratch.

At around that time I noticed the odd bogus character in the output going to a terminal. It happened very rarely and I thought at first it was a problem with the terminals input buffer being overrun or just noise. When I finally realised it was the driver I spent about four hours pulling my hair out before I tracked it down to a reversed increment/mask operation in the interrupt service routine: buffer_pointer &= (BUFFER_SIZE-1);
buffer_pointer++;
which should of course have been
buffer_pointer++;
buffer_pointer &= (BUFFER_SIZE-1);
As the buffer in question was 4k in size it happened very rarely!

After this I spent a lot of time experimenting with the best way to send characters to the transmit buffers on the card. The best results were obtained by sending as many characters as possible to one port and then moving on to the next. I was mainly concerned with performance. If people spend upwards of 400 pounds on high performance serial hardware they deserve the best from the device driver too.

Most applications where high data rates are used (SLIP/PPP etc) will put the driver into raw mode. But the card also supports output character cooking (OPOST) for interactive use, which I was keen to implement.

The problem with the Linux tty system is that the line discipline routine calls the device driver rather than using the usual Unix method of having the device driver call the line discipline. The end result is that cooking has already been done by the time the device driver module sees the data.

I have been considering four options to circumvent this problem.

• One, modify the line discipline module to have it check when to do cooking. At the moment of course, the line discipline always does the cooking when required to do so by the settings in termios. The idea is that the driver would set a flag at boot time to tell the line discipline that it is capable of doing it's own cooking when required. When using this method the line discipline module has to be selective because the hardware cannot do all types of output cooking, just ONLCR. So it looks at the termios structure and if ONLCR is set and no other cooking flag is set (like tab expansion) then it would switch itself to raw mode and let the driver, or rather the hardware, do cooking. The disadvantage is that you have to check several flags rather than just one. This will be worse in future if other hardware supports other types of cooking.
• Two, as suggested by Ted T'so (author of the Linux tty subsystem), is similar to the first method. Have all the logic for deciding when and when not to do cooking in the driver with a flag that is checked by the line discipline. One advantage of this method is that it makes the changes to the line discipline code very small as only one flag has to be checked. Only a single check is needed per block of data that the line discipline processes whereas in method one, several checks have to be carried out for each block of data.
• Three, also suggested by Ted T'so, have the driver fool the line discipline into doing raw output. In this method the driver would keep it's own copy of termios data and modify the copy that the line discipline code sees. You have to keep a copy of course in case the user level needs to look at it. This scheme is rather complicated to implement but it does have the advantage of keeping everything in the family, in other words it requires no change at all to the Linux tty sub-system.

  The Linux tty sub-system works very well as it is. It is simple and clean and would become more complicated by having to check for 'intelligent serial card' flags (although these checks can of course be entirely removed by a compile time option for systems not using intelligent serial hardware).

• Four, don't use hardware cooking at all. This isn't as bad as it sounds because for most applications, when the driver is working hardest it will be in raw mode anyway. The only people to lose out would be those using the driver for lots of interactive users. For these applications, after about eight or so ports, performance would fall off dramatically because the system would be processing output data character by character instead of as blocks.

I have distributed the alpha version of the driver with a modified n_tty.c (line discipline module) which implements the first scheme above and was simple to implement as an initial attempt.

The latest news is that Ted Tso is hoping to add support for another intelligent serial device himself, so the kernel will probably see intelligent serial hardware properly supported in future (whatever scheme is decided on!).

I started a new job at the begining of last September and apart from adding support for the EISA version of the card I've not had time to do any more work on the driver since then. I hope to move the driver to Beta status, if all goes well, in the next two months. I've just bought another 32 port SIO system so testing with a large number of ports will also happen soon!

Background Information

Each external module contains either four or eight ports for a maximum of four modules per card or 32 ports. Each port is typically RS-232, configured (unusually) as DCE rather than DTE but modules containing a Centronics type parallel port are also available. It is possible to have a maximum of four cards per system (total 128 ports!) but my Linux driver only supports a single card at present.

I have uploaded the alpha version of the driver to sunsite.unc.edu but it should be available by now at any mirror site as well.

I am quite willing to speak to anyone about the Specialix driver or other intelligent serial card related issues and can be contacted at simonallen@cix.compulink.co.uk

I work as a software development engineer and my current project is the upgrade of the firmware on a multiport X.25 card(!) so perhaps you can understand why I don't really have the time or inclination to work on the Linux driver at the moment!

There are now 4 linux-uk lists, they are

• linux-uk-discuss A general discussion group on Linux.
• linux-uk-help A help and answers group. No query too basic, no flames!!!
• linux-uk-usergroup Setting up and running the usergroup
• linux-uk-all The supergroup - everyone on all the other groups, intended for announcements.

I am NOT saying that you have to be a paid up member of the UKUUG Linux SIG to use these list. We welcome anyone who has an interest in Linux, although I suggest that you will get most out of the list if you actually live in the UK as we hope to be using the list to arrange local events.

HOW TO JOIN

To subscribe send a message to
        <mailbase@mailbase.ac.uk>
with the command
        join linux-uk-usergroup Firstname Lastname
        join linux-uk-discuss Firstname Lastname
        join linux-uk-help Firstname Lastname

If you want to unsubscribe to any of the lists then send the command

        leave <listname>

        suspend mail <listname>

        resume mail <listname>

You can leave the subject of the message blank. Be sure to send the mail from the account that you want to receive mail from the mailing list in (more of this later). You will be emailed confirmation that you have joined the list and also a handy mailbase 'reference card' for you to print out and keep.

THE LISTS

• linux-uk-usergroup
  This is a low volume list that will contain notices of the activities of the Linux SIG of the UK Unix Users Group such as discussion of what is going into the newsletter and organisation of local group meetings.
• linux-uk-discuss
  This is a free for all list for discussing and announcing things of general interest to the UK Linux community (even if they are not members of the Linux SIG). This is expected to be quite a high traffic volume group so you may like to just collect the monthly archives rather than have messages delivered individually.
• linux-uk-help
  This is for people wanting help with linux problems. This has been split from linix-uk-discuss because in the early days of the linux-uk list Linux newcomers needed to know some simple stuff which old hands felt was a waste of their time to receive and read. Everybody please spend at least some time on the help list to share the Linux experience with newcomers!

  Remember you can suspend and then resume membership of any list if you want a break from the message stream!

  THE UKUUG LINUX SIG WWW PAGE

  Point your WWW browser at:
  
  http://web.dcs.bbk.ac.uk/ukuug/sig/pc/linux/home.html

  This will get you the home page of the Linux SIG - this will contain pointers to many interesting things including the Web version of this newsletter!

  ---

  Linux Local Groups Guide

  By Martin Houston (mhouston@mh01.demon.co.uk).

  What is a Linux Local Group?

  The answer is whatever you want to make of it! The basic principal is that it is a mechanism for Linux enthusiasts to meet face to face rather than through the net. The advantages of this are that people can talk about shared problems and interests in a spontaneous way rather than the email based conversations that have gaps of hours or days. Meeting together allows people to actually SEE things being demonstrated rather than having to work just from written instructions.

  A workshop where new users are shown how to do a Linux installation would be a good subject for a first meeting. This assumes somebody has a 'spare' machine that they don't mind being practiced on!

  Where should a group meet?

  The simplest arrangement, for a small group would be to meet in the organiser's home. This has advantages in that it is free, a Linux computer is at hand.

  The disadvantage is that the number of people that anybody would want to have descending on their own homes is strictly limited! Meetings would have to be by prior invitation only. Suddenly having 50 people turn up to your house is a sure way to upset the neighbors.

  If a local group has reached more than a handful of members then a meeting place will have to be found. Look at Pub function rooms, civic centres, church halls, public libraries and similar places. It may be that local companies or institutions with an interest in Linux may have a meeting room that they would be willing to host a Linux group meeting in? Local universities or colleges are usually hotbeds of Linux interest so may be worth approaching for facilities.

  Unless one of the group members has such a resource that they would be willing to share hiring a room will incur a cost which should be passed on to those attending. Twenty pound room hire translates to a pound or two per person - a small price to pay for an evening of being able to chat about Linux!

  There is an important point to note about liability and insurance if your group is meeting in a public place. Contact UKUUG before any such event can be organised so that appropriate insurance can be taken out. It is usually a stipulation of such insurance that the Linux group is run as a club with only members and signed in guests present rather than a general public event. This requirement also gives us the chance to sign up new members at meetings so they can bring their own guests next time.

  Other Advantages of Face to Face meetings

  One of the things that the Home Computer clubs of the early eighties were famous for was the high degree of exchange (piracy) of software that went on. In the enlightened Linux world such exchange is no longer wrong! Sharing packages that you have found to work well with others - along with a demo of why they are useful to you is the sort of mutual help that will make the Linux clubs so useful.

  Concentrating Linux enthusiasts together in one place also means that the sort of Linux merchandise such as tee shirts and bumper stickers that would not be worth the trouble of personally importing from the states or wherever becomes viable and fun to do as a group. Want printed documentation? If your group are all interested in a new package why not get a local printer to run off some nicely bound copies of the documentation.

  One of the things that would be useful to do at a local level is to appoint somebody as custodian of a complete set of up to date Linux installation floppys so that they can be borrowed so that anyone joining the group can try Linux on a machine for no cost at all - apart from a deposit on the disks maybe!

  The secret to making a success of Linux is to keep costs down by avoiding waste. A big Linux community doing some things on a distributed local level but getting together as a national or even international organisation to deal with the big things.

  ---

  Linux Archive CDs

  by Martin Houston One of the things that the UKUUG has the capability to do, and would be very useful for Linux users is to produce a CD archive of all the Usenet user groups of relevance to Linux together with the UK's own linux mailbase archives.

  Such a collection would mean that a this raw flow of Linux material could be browsed offline to find things that take your interest or serve your needs. This without having to spend hours on the telephone or clogging up precious hard disk space.

  Such a collection would be complementary to the normal Linux CDs that are dumps of archives rather than just the message flow.

  One aspect of CD production is that it can only be done at the sort of cost that brings CDs to the front of magazines when the production run is large. If the Linux community in large numbers say that "yes I want the convenience of getting net resources archived on CD" then we can do it. I envisage that provided that we can get the critical mass of interested people Linux CDs can be produced for as little as 10 pounds a time - or less if we can get together on a more efficient method of distribution than individual mailing of copies.

  This is one area where local Linux groups might become useful - they can collect orders for and then distribute regular archive CDs of what the Linux world as a whole and the UK community in particular are up to. I need feedback - suggestions of things that you would like to see on the CD. For starters all the archives of the linux-uk mailbase groups will be there along with web copies of "linux@UK" as they are produced. Developmental snapshots of any Linux projects that the UK community are working on would also be a good idea.

  The net distribution of slackware and maybe some other popular distributions and Linux add-ons such as Wine, Mosaic and DOS-EMU so that we can easily choose to update on a regular basis if we want to.

  What I am aiming at is a CD that is updated quarterly, bi-monthly or even monthly if there is the demand that eliminates the need to fetch large amounts of news and data from the Linux net servers. For people who work within a local Linux group this should be a rich resource that maybe eliminates the need for a personal net connection. Even for those with a net connection it means that there is less need to download and use up disk space for things if they come to you on CD.

  To take an example of what I mean - You download and compile the latest version of a BIG package like NCSA Mosaic from the net - it works fine, what do you do with all the source files? If it is one of the things tracked on the CD then when your next CD arrives, check that the CD version is the same - or better :) than the one you fetched from the net. If it is then you can free the hard disk space taken by source of Linux tools that are not currently under development. If you have made changes to your version then these can be compactly stored as a patch file on a floppy disk (along with a note as to where the related CD archive can be found).

  What I need from linux@UK readers is some feedback as to the takeup of such a CD. It is only if large numbers of people what the convenience of the archive that the price of producing one will be low enough to make it attractive. Like any publication it's a chicken and egg situation. As Linux and the Internet archives are in a state of flux we need to know that enough of you want convenient CD snapshots to be able to make them in quantity. Also your suggestions of what is good enough to go onto the CD would also be welcome. Just the list of files that is on Sunsite Northern Europe is nearly 10MB of data COMPRESSED!. Finding the best 600MB of material of interest to Linux users from this is best done co-operativley. Also don't just recommend something - write a short piece for us about why you have recommended it. In this way we can share the good news of what is in the archives to be used. Knowledge is Power The point of this user group is to share experiences with using what is out there. I hope that by the next issue we shall be able to announce the availability of the first of our CDs. If you know of something that deserves to go on the archive let me know.

  ---

  Linux Groups Around the World

  by Martin Houston

  We would like to talk to Linux groups around the world. If you are running such a group then please send me details of it for inclusion in this section. To start the ball rolling some information from the Californian Linux Users Group!

  Cal Linux Users Group Meeting, Berkeley, CA, USA

  If you're a Linux user or want to become one, CalLUG is the place for you. Whether you're interested in getting started, surfing the net, system administration, doing your class assignments at home, or heavy-duty kernel or X hacking, you need no longer be alone. CalLUG hopes to be a resource for new users, a forum for the exchange of ideas pertaining to Linux, and a contributing member of the Linux community. At the meeting, we will discuss topics of interest to all users, talk about what the future role of CalLUG should be, and distribute resource lists of local experts in case you have questions or need help. Please join us on Thursday and contribute to this ground-breaking organization.

  If you have any questions, send mail to callug-board@amacrine.berkeley.edu.

  Paul Eastham Chairman, Cal Linux Users Group

  (excerpt from mail from Paul Eastman to Martin Houston...)

  Hi, thanks for the good wishes. Looks like you're pretty far along...our first official meeting is on this Thursday.

  I've been in touch with the leader of the UCLA Linux users group (UCLinux) and we have been exchanging ideas as well. Perhaps at some point we could create a mail alias for presidents/chairmen of Linux Groups throughout the world to exchange ideas and reports on how things are going.

  Here's a few things we've got planned so far:

  • A "home system" sacrificed to one of the campus machine rooms where everyone can log in, hang out, chat, and deposit/retrieve linux-related files.
  • A web site (under development at http://amacrine.berkeley.edu/callug/) for members' home pages, info, etc.
  • A reference list of local experts (CalLUG members) on topics from installation to hard disk partitioning to kernel development.
  • Resources and assistance for people seeking to use Linux with their CS coursework (much of the software for Cal CS classes has been ported to Linux)
  I'll let you know how it goes.
  
  Paul
  Chairman, CalLUG
  

  ---

  Linux kernelitus?

  In this issue, Lasermoon takes a somewhat irreverant look at Linux :

  It's out. Kernel release 1.1.99z-1 with support for the Aztec Sound CD combined Graphics adapter and a beta version of the Spectrum 128K games port. Available on all good ftp sites. OK, so this is a spoof, but it is not that far from the truth of current Linux development. There seems to be a new kernel release every day. More on a good(??) day. And of course, they are all fully tested so there are no problems or upgrade difficulties. And you get the latest news of the latest kernel release on the Internet. Not from any authoratitve source, but from respondents on a News group. OK, so this is what has contributed to Linux's break-neck development speed and you will doubtless find descriptions of this elsewhere. But back to the kernels. A good example of the problems caused by Internet gossip came today with a conversation with a customer who had a PC from a well know manufacturer with a CD-ROM which turned out to be an incompatible with a well-known distribution. The customer went back to the supplier who assured him that the CD WAS compatible with Linux, Kernel 1.1.85 to be exact. They knew this from a reply received from a posting to the Internet. The customer then asked, quite reasonably, why the CD he had did not have 1.1.85 on it and what could be done. This raises some interesting issues :

  1. The "source" on the net was little more than gossip, but the hardware vendor (a BIG name as well) used it as a sales feature.
  2. Nobody considered what was wrong with 1.1.85 or what it had been tested against (if anything).
  3. Everyone assumed that the reply from the Internet was authoritative and accurate.
  It turned out that there was someone on the net who had tested it and it worked OK. For him. On one PC. In the USA. But he had IDE drives and not SCSI. And no sound-card or ethernet either.

  But isn't this part of the fun of a new OS?? I mean, you are at the cutting edge of development and there are bound to be a few problems. Like the major Linux releases with the following howler :

          The X based version of Emacs takes 10 minutes to load if an
          Ethernet card is *not* present.
  

  And this is on a 1.1.59 kernel. Why wasn't this found earlier? Like all the little irritations in the manual pages, and the oddities in the basic utilities (try the following command : od -N 200 ) the hack to gtbl and the total lack of locale support. These are not fun. They are not exciting. And there is nothing to test them against as the underlying kernel is changing far, far to fast.

  A recent Linux conference in the USA had a debate with a theme of "the best Linux kernel". Guess what - the 1.1.18 (this is not a typo) got the vote from the assembled mass of developers. The new kernels suffer from bloat. Creeping Featurism (the new 1.1.99z-22 kernel supports the new ACME TIG-welding interface) seems to have taken over. New kernels are slower and more cumbersome as a result. And they are more bug-prone (any serious software developer is aware that bugs increase exponentially with program size and complexity) and require more testing. Testing which is simply not possible with the new kernel releases. The trouble with the new kernels is that they contain too much code, many features being relevant to a small section of the user base. It's not bad, it just needs a little control.

  There is an intermittent thread on the News Groups about Linux not being taken seriously by the ISV's (Independent Software Vendors) and the lack of applications for Linux. Of course Linux is not taken seriously! It never will be when the underlying operating system kernel is changing so fast that accurate testing is both impossible and pointless. A typical application has to be tested against around 5 (deliberately different) distributions, all with different kernels and most with both X11R5 and X11R6. Even if an ISV managed the herculean task of testing against all of them they would still :

  • - Lag far behind the latest kernel (now 1.1.99z-2 :-)
  • - Have to cope with the customer changing kernels and then complaining that the application does not work properly.
  • - Have some Internet-weenie starting a rumour that the 1.1.99z-4 kernel is better than the 1.1.99z-6 kernel....
  Considering the huge manpower and $$$ costs involved in testing any application release, the ISV's simply wont port applications to Linux whilst it is changing more than a kernel release every 6 weeks. Sure, there are some GREAT application packages for Linux. But do we find Frame, Informix, Oracle, Teamwork, Stock control, Accounts, payroll, inventory management, DTP etc etc??? All things that would increase Linux's standing in the market and also its user-base. The other OS suppliers (we all know who they are) regard Linux as (at best) an interesting diversion, but "If you want to run Applications you have to use XXXX" where XXX is the OS of your choice (OSF/1, Solaris, UNIXWare, SCO etc...)

  The message here is more than just cynical ramblings. Whilst Linux IS the probably the fastest developing piece of software on the planet, the very nature of its development could be its undoing.

  Fortunately things are not all doom (the 1.1.99z-45 kernel can support multi-user Doom in 1280x2048 resolution). 1995 will see a number of new Linux releases aimed at the Commercial sector and ELF format will become a standard part. The Commercial Linux distributions will be aimed squarely at people who want to do things *with* Linux rather than *to* Linux. They will provide the ISV's with a slow-changing base onto which they can port their applications. Spec 1170 certification for such distributions would represent a huge shake-up in the UNIX industry - and it's likely to happen this year! The inevitable outcry from the Free Software purists that commercialisation of Linux is bad and that a 2 tier Linux will result is almost irrelevant. So what if there is a 2 or even 3 tier Linux development path? It would represent a win-win situation where the Commercial distributions allowing larger numbers of applications to be ported with a commensurate increase in the user base and allowing better testing. And the Internet development could continue (the 1.1.99-47 kernel having broken the ear-wax melter, does have the 45rpm single player fixed) as normal. Nobody looses. Everyone gains.

  Which is what makes Linux the best!

  Next Issue (with the Ed's permission) :

  A look at the newest distributions and products, a delve into the murky depths of setting up an ftp server and another tongue-in-cheek investigation into the latest gossip (maybe we'll have a 1.2 kernel by then!)

  And who are Lasermoon? We are UK specialists in UNIX Freeware, Linux and BSD distributing products from Infomagic, Morse Telecom, Yggdrasil, PTF, GUI Corp and many, many more. We have a free support service for our customers and an ftp site on ftp.lasermoon.co.uk Feel free to contact us on info@lasermoon.co.uk If you'd like to discuss this article, please send mail to sig-article@lasermoon.co.uk.
  

  Lasermoon Ltd, 2a Beaconsfield Road, Fareham, Hants, England. PO16 0QB, Voice +44 (0) 329 826444 Fax: +44 (0) 329 825936

  ---

  Your Editor: Who am I and why am I here?

  This is me with my daughter Felicity who at three and a half already has a keen interest in computers!. I have been interested in computers ever since University. I first encountered Unix in 1981 on a PDP11-70 computer with core memory! That was at Queen Mary College, London. I only learned afterwards that it was the first Unix system to be set up outside America.

  I spent my early working life as a systems programmer for Microsoft Xenix and then SCO Unix operating systems. I have had a keen interest in C programming and helped found the C Users Group (UK) - now the Association of C and C++ Users. In the early days of the CUG I was Chair and Software Librarian. As early as 1988 the CUG had an interest in Free Software Foundation projects and carried several of its products including Emacs and Bison in the source code library. I have had an enduring interest in the Free Software foundation for many years - Linux seems a natural home.

  ---

  UKUUG Linux SIG Membership Application

  Scale of charges

  • Existing UKUUG members NO CHARGE
  • Existing ACCU members 15 GBP
  • Neither of the above 20 GBP
  Payment for the 1995 membership year may be made in one of two ways:
  1. By UK Cheque or Bankers Draft, made payable UKUUG and drawn on a UK bank.
  2. By Visa, Access, Eurocard or Mastercard
     
     |_| by VISA
     
     |_| by ACCESS/EUROCARD/MASTER CARD
     

  Name as it appears on the card (block capitals)

  ---

  Address
  

  ---

  
  

  ---

  Postcode

  ---

  Email (if any)

  ---

  Daytime Tel

  ---

  Date of Expiry

  ---

  Card Account No.

  ---

  Signed

  ---

  Date

  ---

  Send to: Linux/UK SIG
  UKUUG Secretariat
  Owles Hall, Buntingford
  Herts, SG9 9PL
  

  ---

  That's All Folks!

  Thats all for this issue of Linux@UK. Hope to see you all again in a couple of months!
  Martin Houston