[UKUUG Logo] Copyright © 1995-2004 UKUUG Ltd


Table of Contents Next


Editor's Column

(Susan Small)

Sue As I am editing this edition of the newsletter on a beautiful autumnal day (1 October), I know what a treat you are in for when your paper copy hits the mat.

Richard Ibbotson's report of our Linux Conference in July gives a real flavour of the event (almost as good as being there!); Bar38 must have wondered what hit them. Sebastian Rahtz' trip to the Open Source Software Convention in California is well documented; having attended a considerable number of conferences myself (on both computing and other subjects) he certainly reminded me of the "emotional and physical exhaustion" that can set in by the third day!

A number of our reviewers brought a smile to my face too, waxing lyrical about the books we sent them (or in one case, not). If they are to be believed, a number of these books will change your life. Let me know if they do. To celebrate a new "Illiad" book, there's a sample of the humour on the letters page - along with a job advert for Web developers.

You can also read about how the Americans are supposedly working on speeding up the Internet (this is certainly good news for online shopping:-) and, more worryingly, about the FBI's Carnivore PC specially designed for tapping into Internet traffic.

I hope you enjoy Professor Staplefood's Lent Lectures.

Chairman's Report

(Charles Curran)


New Council Members

At the AGM, the Council gained two new members: Roger Whittaker, who works at SuSE Linux Ltd, and Owen Le Blanc, from the University of Manchester.

Sebastian and Roger Roger joined SuSE as their training consultant at the beginning of this year and is involved in setting up training partnerships, liaison with schools, LUGs, training for resellers, distributors, et al. Prior to that he was deputy head at the Hellenic College, London, where he had worked many years.

Roger is on the right of this picture talking with Sebastian Rahtz.

Miguel and Owen Owen (Audoenus) has been working at the University of Manchester (UMRCC => MCC => MC) for the last 15 years. He originally hails from Louisiana. His academic career stretches through many universities and many subjects (a BA in Classical Languages and Mathematics, a BD in Philosophy and Theology, and an MA and a PhD in Logic). In the computer world he has experience of a wide range of hardware, OSes, and languages; one of his major interests is in the formal specification of languages. Some of you may have encountered Owen in his work on Linux's fdisk, on the MCC Interim Linux distribution ('92-'96), or perhaps, more recently, at UKUUG's Linux2000 conference, where he gave a talk on the CODA network file system.

Owen is on the right of this picture talking with Miguel de Icaza.

News from Owles Hall

(Jane Morrison)


I am pleased to announce that at the UKUUG Annual General Meeting on 21st September 2 members were nominated and duly appointed to Council:
Roger Whittaker (SuSe Linux Ltd) and Owen LeBlanc (University of Manchester).
We are looking forward to them joining Council and would like to welcome them aboard.

We would also like to thank Drew Durkin - UKUUG Treasurer, who stood down from Council at the AGM. Drew (from Newcastle) has been on Council for the last year and although has stood down due to work commitments, hopes to continue to assist with the February 2001 Conference which will be held in Newcastle. The UKUUG AGM minutes will be circulated to all members very shortly and will also appear on the web site. I can confirm that the first meeting on the new Council will be held on 24th October.

By the time you actually receive this Newsletter it may be too late to book a place on the forthcoming Tutorial - DNS Administration by Jim Reid of Nominum (5th October). To date we have 23 bookings and it looks as if the day will be a great success.

We are hoping to organise more Tutorials in the next few months - perhaps you can give us some topics they you would find of interest.

The recent event in July - The Linux Developers Conference was a huge success with over 100 delegates. Leading speakers from around the world attended to make this the 3rd annual Linux Conference for the UKUUG. The event made a very small profit which proves that we had calculated our budget correctly. See Richard Ibbotson's write up on the event in this Newsletter.

Plans a now afoot for the 2001 Developers Conference which may be held in Manchester. Watch the web site for more details.

The Newcastle Conference mentioned above is planned for 8th & 9th February 2001. The venue will be the University of Newcastle and a suitable city centre hotel will be identified as the Conference hotel. You should find a Call for Participation flyer enclosed with this mailing that provides more details. Please put the dates in your diary now.

I have been checking our membership database and I have quite a few members email addresses missing and some that look conspicuously wrong. Please see the 'Membership Data Sheet' enclosed and if you could spare a few minutes to complete and return to me I would be very grateful. This data sheet can also be found on our web site - www.ukuug.org/membership.

We are looking for members who may be interested in joining the book review team and who would be willing to write regular articles and columns for the news@UK - if you are interested please let me know - office@ukuug.org.

Copy date for the next issue Newsletter (December issue) is 24th November.


(Kieran Barry)

YAPC stands for "Yet another Perl Conference." The first two were organised by Kevin Lenzo of Carnegie Mellon University, to cater for the audience which might not afford O'Reilly's The Perl Conference, the flagship event in the Perl world. Incidentally, O'Reilly have responded by providing rather generous sponsorship to YAPC events, and this one in particular.

The aim has been to provide a low cost, low frills, technical conference organised by and for the Perl community. From Friday 22 to Sunday 24 September, the ICA in London hosted the first YAPC::Europe. At £40 per head, the organisers underestimated demand significantly and could possibly have attracted twice the attendance.

I only cover those talks I attended.

Friday was tutorial-directed. In the morning, Stas Bekman, a key developer of mod_perl, spoke on how to set up and tune an Apache Server with mod_perl, as well as mentioning a number of traps. His key recommendation was that a proxy server such as squid be used so that a perl-enabled server could run at peak speed. This would give the highest through-put per unit memory used.

In the afternoon, Michael Schwern presented "Writing Solid Perl". While dogged by technical problems early on, he tried to demonstrate his development procedure for writing modules in Perl.

He recommended use of a number of tools like CVS, h2xs (a header file converter), Makemaker and POD (Perl's Plain Old Documentation) to generate documentation and test suites as the code is written.

The second half of the talk turned into a discussion of optimisation. He strongly urged the audience to be reticent about this, unless absolutely essential.

Michael was suffering visibly from jet-lag and did a sterling job in face of adversity.

Saturday morning kicked off with a description of YAS (Yet Another Society), from Kevin Lenzo. This is designed to aid organisers setting up grass-roots, low-cost technical conferences.

The Keynote was given by Simon Cozens. His theme was a comparison between Open Source Software and Art. It was certainly more stimulating than the standard keynote fare; as a self-trained philistine, even I wound up in a serious discussion about what constitutes art.

Honza Pazdziora next gave a talk about what he calls Docserver. This is a client-server setup written in Perl to get MS Office programs to translate documents in proprietary formats into text, HTML or postscript, and serve these to clients on Unix boxes.

brian d foy (sic) then gave an overview of plans for Perl 6. The 2-month Request For Comment period comes to an end on 30th September. On 14th October, Larry Wall will respond to all Perl 6 RFCs, and define what will go into Perl 6. An Alpha release should be available next year around the time of TPC 5.0. Perl 5 will also be supported for some time to come.

At Lunchtime, Nat Torkington from O'Reilly held a pickup meeting to discuss Perl 6. Nat is project manager for the Perl 6 development effort. He commented on the RFC process currently underway (roughly: "People are making suggestions for language changes based on problems they have, but the RFC structure means that they have to leave out the problem from the RFC").

After lunch, we had a session of 5 minute talks:

Following this, Charlie Stross gave a talk on Netserver::Generic. This allows the programmer to write a TCP server-program in a stress-free way. It provide a file handle to read and write from. A key problem is that the Perl <> operator, which reads to the end of line, can leave a server open to denial of service attack based on extremely long lines.

Dave Cross gave a talk retitled Perl is Boring (advertised as Datamunging in Perl. He renamed it after it started.) His points seemed to be that Perl is ideal for data translation, which is less sexy than some of its web uses, but that it was important to approach such problems systematically. He has written a book on the subject due out shortly.

To round off Saturday's program, Piers Cawley hosted 12 Step Perl, a confessional session based on the structure of an AA meeting. Sadly, I seem to have lost my notes :)

Saturday evening saw a large pub crawl through Southwark.

The only talk on Sunday I saw was Kevin Lenzo talking about a speech generation package based on Festival called Festvox.

To round off proceedings, Greg McCarroll auctioned of a number of books donated by O'Reilly and Manning.

All in all, I enjoyed the experience. Thanks should go to Leon Brocard, Greg McCarroll, Dave Cross, Kevin Lenzo and a cast of thousands who helped to make it happen.

Sophos spreading FUD about Linux Viruses

(Eddie Bleasdale)

In Computer Weekly, 29th June 2000, a letter from Sophos Anti-Virus Laboratories claimed that viruses exist for Linux. Eddie Bleasdale, director of netproject, has asked Sophos to demonstrate these viruses on a Linux computer that netproject supplies. Sophos has refused.

'We have been working in the Unix area for over 20 years.' said Eddie Bleasdale 'During this time we have never encountered a Unix or Linux virus nor have heard of any organisation that has been infected by a Unix / Linux virus. We need to stop the fear uncertainty and doubt that the anti virus companies are trying to create around Linux.'

Eddie Bleasdale does not doubt that viruses can be written for any operating system. What is different about Linux, compared with Windows, is that there is no need for anti virus software because controls exist to ensure that only authorised software runs on a correctly configured and administered Linux computer. These controls do not exist for Windows. 'We believe that Linux is pretty much bullet proof and if Sophos are able to infect a well configured Linux box then it will uncover an implementation defect rather than a design flaw. Anti virus software simply treats the symptoms and does not address the fundamental design weaknesses that allow viruses.'

The challenge to Sophos is to send an email with attachments. These will be read and attachments opened. There will be no anti virus software involved in this demonstration. The demonstration can take place in the laboratories of Sophos.

'We believe that this is a fair challenge and it is not one that Microsoft would be prepared to offer. Viruses are a fact of life with Windows because of the design defects and the complacent attitude Microsoft has to security.' said Eddie Bleasdale.

So far Sophos, despite having claimed that viruses exist for Linux, has refused to demonstrate them on any Linux computer that Sophos has not configured.

The response received from Sophos is:

I don't have any response other than that which I have already given you.

I'll give the same response to any journalists who might call me up.

I don't have any more to say on the matter. I think it will be a waste of our mutual time if you email/phone Sophos on this matter again.


Graham Cluley, Head of Corporate Communications, Sophos Anti-Virus
email: gcluley@sophos.com                    http://www.sophos.com
US Support: +1 888 SOPHOS 9            UK Support: +44 1235 559933

Report on the Conference on Open Source in Education

(Mike Clinch)
UKUUG Council Member for School Focus

Dr. Malcolm Herbert of BECTA, gave the Seminar Introduction. The intention was to help BECTA to develop plans and strategies for the future. This was BECTA's first Technical R & D seminar. Existing research in BECTA, over the last 12 months had indicated there was no NGfL approved suppliers of Open source.

The intended outcome of the seminar was to:

  1. Clarify BECTA's position on Open Source
  2. Determine any future action
  3. Improve the procurement process allowing for the inclusion of Open Source

There were wider issues to be addressed. There needed to be a cultural change in school IT, there was a need for learning centres, and there was the difference in the needs of schools from those of industry as a whole.

Malcolm Herbert then introduced the first case study.

Dr. Martin Williams, Powys LEA, IT Advisor for Schools

Martin presented a case study of Powys LEA and Open Source. Powys was described as the largest of the local authorities in Wales. It is 100 miles from north to south with a very large dispersed rural element. A considerable advantage had been the close working relationship between the different departments and people involved in the project. The network was to provide regional information. It started in 1992 with the Rural Wales Network. Access to the internet was added in 1995. By 1997, Business and Tourist Information Centres had joined the network as had Youth Facilities, Libraries, Museums, Telecentres, and Voluntary Sector Organisations.

NGfL issues required sustainabililty, allowing high capital cost to be balanced by low maintenance costs. For schools it was decided that email for all was required and that a web server be provided in each primary school. The network was provided as a "figure of eight" based on 7 towns. This approach required high costs to implement but allowed extensions to be made at lower cost. The provision of the network infrastructure became interesting, especially when extending it to the more remote schools with perhaps as few as 28 pupils.

Financial restraints left little for the servers after the network was in place. The need for servers were a problem until Open Source software was tried. Servers were bought at some £240 each against some £1000 for an alternative server. Cost reductions could be made since the servers were to function without VDUs or keyboards. Administration tools for teachers had to be crafted to simplify user management. This is provided through a web-based interface. Tools like this filled in those few gaps found in the Open Source software. All system administration is done by the central IT group with server configurations held on a central database. Usage graphs and logs for all nodes have helped to identify bottle necks and so improve the network functionality. The benefits of this approach were Server costs of £240 rather than £1000 for the alternative. Stunning server reliability and A1 support.

See http://www.ose.org.uk/conference for a PDF of the slides. Malcolm introduced the next speaker.

Damian Counsell of the Institute of Cancer Research

Damian gave a presentation on the Human Genome Project (HuGeP). A very interesting overview of DNA was given. Having been given the simple explanation of what DNA is, it was then explained why it is desirable to learn the sequences of it. The Sanger Centre, Cambridge is the UK centre for the HuGEP. Workers here have co-operated with other centres with the sequencing work. Damian explained how over sampling was used to extract multiple short sequences of DNA and then raw computing power was used to fit these pieces into place.

This approach was used to try to speed up the process of determining the DNA sequence. There is an enormous number of short sequences, many with overlapping sections, to be fitted into the complete DNA sequence. It becomes clear why there is a need for a reliable processing power to match fragments and piece together the complete sequence. This raw processing power was provided in part by a large Linux cluster. This cluster was reliable enough to run for days on end, matching bits of DNA sequences. There are tools to assist in identifying DNA sequences, Emboss and Ensembl. Ensembl is an Open Source tool ( http://www.ensembl.org), and is available for download.

The sequences obtained have now been published on the Internet, as open source, for anyone to download. You may, if you wish, download all the sources and use your spare cycles to do your own matching! However, Gene Patents are coming under tighter scrutiny, so do not hope to make any quick money by guessing what small sequences may do. It was commented that Open Source software was used and reflected the ethos of Open Source DNA Sequence. It should be remembered, Open Source in science allows for peer review.

After Lunch there were discussion forums on Open Source and Licensing.

Mrs Farrington of the DfEE was interested in information and advise on using an Open Source type license for bespoke software commissioned by DfEE. She was interested in the release of DfEE-owned software to clients who would be free to have their own changes made to the source, at their own cost. The modified source would then have to be passed back to the DfEE. It was pointed out that any changes would have to be additional to the original core source or DfEE could end up with two or more sources which were not compatible.

Mrs. Farrington showed concern that there were no case studies of the advantage of Open Source in use. Unless these case studies could be provided, there would be problems in determining that Open Source was an advantageous purchase using tax payers' money.

It is a pity that there are many cases where Open Source has been an advantage, but are not available as case studies.

The seminar was closed by Malcolm Herbert. He gave out information about future developments and forthcoming seminars.

Mozilla and Netscape 6

(David Hallowell)

They're have been a lot of reports recently about Mozilla and its commercial spin-off Netscape 6, there's a lot of people who think the project is dead and even more people who complain that Netscape are spending too much time on gimmicks rather than producing a fast and stable web browser. In this article I'd like to present my take on the Mozilla project and give details on how you can help out with the project or at least make sure that your web pages work correctly under Mozilla.

Why has it taken so long?

Probably the most frequently asked question about Mozilla is why has the project taken so long. Netscape released the source to what was going to be Netscape Navigator 5 (the original source release contained Navigator and Composer only minus software they had licenced from third parties including Java and crypto) in April 1998. The orignal plan was to release a Netscape 5 beta by the end of 1998 based on this code. However, as work was progressing on this browser they realised that this code had outgrown its usefulness and the only way they could produce a high performance standards compliant browser was to scrap just about all the original Netscape code and start again with a ground-up rewrite. The main component of this new browser would be their new layout engine ( http://www.mozilla.org/newlayout/ - now often referred to as Gecko). The browser would be written with as much cross-platform code as possible to ease the porting to other platforms and help ensure that pages would display almost identically on each platform (compare with the fact that IE 5 for MacOS has better standards support than both IE 5 and 5.5 for Windows, meaning webmasters may have to detect both version and platform if delievering dynamic pages). Although it probably would have been quicker producing a native browser, this short term vision didn't meet with the long term objectives of the project. However at this time of writing we are getting to a stage where Mozilla is fairly stable and is usable for most people. There's still some major bugs to be fixed both for stability and performance reasons but I use Mozilla day to day as my primary web browser and I'm even using the HTML composer to write this article.

Why can't we just have a browser and only a browser?

Netscape want Netscape 6 to be the replacement for the current Netscape Communicator 4.x series and therefore are paying developers to work on ensuring Mozilla (and therefore Netscape 6) has the features that people are expecting. Many people use Communicator to read mail and news and perhaps to create simple documents. Therefore Netscape put people to work in these areas, Netscape have always had teams of people dedicated to work in different areas, so it's not as if someone working on composer or mail/news is going to detract from the work done on the browser. Netscape only pay people to work on the browser, mail/news, composer and in Netscape 6 the AOL Instant Messenger. To simplify the process of creating cross-platform user interfaces the Mozilla developers developed XUL (XML-based User-interface Language) which is what the user interfaces for all Mozilla components are written in. A by-product of this means that it's very easy to produce skins for Mozilla using Cascading Style Sheets (CSS). Because of the ease in creating cross-platform user interfaces, other people have contributed extensions to Mozilla. For example, you've got ChatZilla which is an IRC client written in XUL and JavaScript ( http://www.mozilla.org/projects/rt-messaging/chatzilla/), an XML based replacement for a UNIX xterm ( http://www.xmlterm.org/), and some other Mozilla-based applications can be found at http://www.mozdev.org/ - the important thing to remember is that these projects are third-party projects, they don't increase the bloat in Mozilla and aren't sacrificing any Netscape developer's time in working on the key components in Mozilla. The foundation that Mozilla has provided has created a massive interest in using Mozilla as a platform for writing Internet-enabled cross-platform applications rather than just being a web browser.

If you just want a web browser then it is possible to install Mozilla with just the Navigator component. However if all you want is the rendering engine (Gecko) with a native interface (rather than an XUL one) then check out Galeon for Linux ( http://galeon.sourceforge.net/) and K-meleon for Windows ( http://www.k-meleon.org/). Particularly on slower machines you may benefit from the performance gains from using a native interface rather than an XUL based one.

So, when are we going to expect a release?

That depends whether you mean Netscape 6 or Mozilla 1.0? The 1.0 release is expected to come after the Netscape 6 launch, with the Netscape 6 launch corresponding to Mozilla 0.9. The exact details of this can be found in the Mozilla Roadmap ( http://www.mozilla.org/roadmap.html), so I won't go into them here, but expect Netscape 6 out by the end of the year and Mozilla 1.0 to be out in 2001. However, as this is an open source project with nightly builds and regular milestones available, you can start using it when it's stable enough for you needs. Personally I'd stick with the Mozilla.org builds. I've been fairly disappointed with the Netscape betas, even though the Mozilla milestone releases they've been based on have been reasonable, the Netscape builds don't offer anything useful (apart from AOL IM if you use that) and tend to come crammed with extra AOL/Netscape advertising. The Mozilla builds don't have this problem and therefore I prefer using them.

Appearance is everything

As I mentioned earlier the way the Mozilla user interfaces are written means that it's easy to skin the browser using mostly CSS; the various collections of skins available for Mozilla are known as themes. Mozilla currently has three themes available by default. The classic theme is now the default for Mozilla and tries to resemble a native application as close as possible with native looking widgets and it also has the same icons as Netscape 4.x for familiarity; the Modern theme will probably be the default theme for Netscape 6, and Blue is the old theme that has icons that look like those on Netscape's homepage and was the default in older Mozilla releases and previous Netscape 6 betas. This theme seems to be here for historical value only and work is being concentrated on the new modern and classic themes. It is possible to download and install your own themes for the browser.

Downloading and testing

You can download the latest releases of Mozilla from http://www.mozilla.org/. The milestone releases are builds created at the end of each milestone and have had some testing conducted on them to ensure that they should run on most of the supported platforms without too many problems. If you'd like to live more cutting edge then you should download a nightly build. These are generated automatically at the end of each day and are therefore not guaranteed to work; sometimes a nightly build will be of excellent quality and the next day it may crash on startup. If you've got a fast Internet connection then use the nightly builds, it's very rare that you'll get a build that's bad these days, but keep a copy of the previous build so that you can use that if there is a problem with the most recent build. Installing Mozilla is easy, particularly if you're using a build that doesn't come with an installer! If you're using a build that doesn't come with an installer just untar/unzip it into a directory and run mozilla (mozilla.exe on Windows). If you previously ran an older version of Mozilla, you're strongly advised to delete the ~/.mozilla directory if running in UNIX or c:\windows\mozregistry.dat in Windows (or c:\winnt\mozregistry.dat on NT). The reason you should do this is because the format of these has changed slightly in newer releases of Mozilla so it may cause problems if older versions are around.

If you find a bug in Mozilla, it's recommended that you read the bug writing guidelines ( http://www.mozilla.org/quality/bug-writing-guidelines.html) before submitting a bug report as this will help ensure that any bugs that you report are likely to be useful and relavant to whoever is going to fix the problem. There's a lot of badly written bugs appearing in BugZilla (the Mozilla bug database) lately, and making sure that your bug reports are well-written and relevant will be a great help.

If you'd like to get involved with the Mozilla project, then you can find some useful starting points at http://www.mozilla.org/get-involved.html. As well as coders, they need people to verify bug reports, produce test cases, and test on the less popular platforms. There's an increasing number of non-Netscape employees contributing to Mozilla, so if this project interests you then do not hesitate to get involved.

Why won't my web page work under Mozilla?

The most likely reason is that your page is using something that was in Netscape 4.x that doesn't fit in with the published standards such as the <LAYER> tag or Netscape 4.x's proprietary DOM. Mozilla closely follows the standards laid out by the W3 Consortium ( http://www.w3.org/) so pages that use specific Netscape 4.x or IE features that are not part of the official standards may not work correctly; however the vast majority of sites on the net do work properly under Mozilla. Some useful references for making standards-compliant pages follow below. Making Mozilla standards-compliant is a good thing; in the short term it'll break a few pages, but in the long term it'll encourage web developers to write to the standards, meaning that people should never have to be tied to one OS to browse certain content. This could be a possibility is IE became too dominant that developers didn't care about other browsers.

Useful Links

Netscape's Mozilla Developer Central (for web developers)
Making the transition from LAYER and document.all to W3C Standards
MozillaZine - Mozilla News
O'Reilly Mozilla Central
NewZilla - The Unofficial Mozilla FAQ
Details of Mozilla.org newsgroups and mailing lists

You can also discuss Mozilla-related matters on IRC. Mozilla has its own IRC server (irc.mozilla.org) with various channels for different areas of mozilla discussion. The recommended channel for general Mozilla discussion is #mozillazine

The search for OGR

(David Hallowell)

Since my last update in the newsletter about our RC5 team, distributed.net has launched the Optimal Golomb Ruler (OGR) project, as oppposed to previous distributed.net challenges which are based on encryption (the current RC5 challenge and the previous DES and CSC challenges) the OGR project is a mathematical project and the results of the OGR effort will have many applications such as sensor placements for X-ray crtstallography and radio astronomy. More details for the uses of OGR are listed at http://n0cgi.distributed.net/faq/cache/134.html.

I won't go into details about what an OGR is but links to all the information that you could ever want to know about this project can be found at the site given above.

If you're currently a member of the UKUUG RC5 team you can help out the OGR project by upgrading your distributed.net client to the latest version available from http://www.distributed.net/ogr/. If you're not currently a member of our team then download the latest client and proceed to the UKUUG RC5/OGR team homepage at http://www.ukuug.org/rc5/ for details of joining the team. Our team's homepage also contains links to the latest stats pages so you can see how we're doing in each of the current ongoing distributed.net projects.

On the RC5 front our position in the stats has dropped a bit since the launch of OGR, this is to be expected as many people will be concentrating on OGR rather than RC5. I agree that our efforts should be concentrated on OGR because of the possible scientific benefits of the project, it may not have the mass appeal of the SETI@home project, http://www.setiathome.com/ but it's probably the most useful thing you can put your CPU idle time to (and the results of OGR may benefit radio astronomy and the SETI project!).

UKUUG Linux Developers Conference

(Richard Ibbotson)

The UKUUG Linux Developers Conference took place from the 7th to the 9th of July at Hammersmith in London. The venue was one of the lecture theatres at Charing Cross hospital. This might not sound too fashionable, but the truth is that the conference was well attended and everyone said that they enjoyed it very much. Many were asking when the next one takes place. It was organised by the UK Unix Users Group and sponsored by SuSE Ltd and Sistina Software.

Friday morning saw the VA Linux team kick off with a talk about Source Forge. Tony "fusion 94" Guntharp gave a superb presentation about Source Forge projects and where all of their projects have come from and are going to. He's one of the original developers and knows most of it backwards. As he explained, we have to find a way to be more organised and more effective in our management of open source software and the general management needs to be improved and will be improved as time goes on. Tony continues to lead and manage the Source Forge team. Sebastian Rahtz works at Oxford University. He came along next and gave us the full works about XML and the documentation associated with it. Also how TEX and LaTEX could be made to work with XML. One of his more humorous comments was "what use could this be to anyone?" The crowd responded with the expected burst of laughter. Sebastian has been heavily involved in TEX for the last fifteen years. Those of us who weren't too sure about hacking XML code learned something from his lecture.

It was when Miguel de Icaza started his talk that things began to liven up a bit more. GNOME is well known by most people who have used Unix systems. As Miguel rightly explained, the development of the Unix operating system has stagnated over the years and some new input is required. GNOME is one of the projects that is in progress that will change the future of Unix. In fact many more people who would have used some other operating system have switched to Unix because of the GNOME and KDE projects. Miguel gave us a very thorough explanation of the GNOME project and he even showed us how to produce the NT4 blue screen of death with the use of Gtk code. He does have quite a sense of humour. This particular exploit did raise a giggle or two. Miguel is the CTO and Chairman of Helixcode at Cambridge, Massachusetts.

Next up was Rik van Rial. Rik writes kernel code for Conectiva in Brazil. His talk about memory management was well received by the audience. His own comment was that he thought that his presentation was boring. Well, perhaps it wasn't anything like as bad as that? He gave us a long and interesting insight into his work and possible future changes that he thinks will take place. Not many of us believed that memory management needed so much hard work but we were more informed after his talk. He also discussed the VM changes in kernel 2.4.

After Rik we were treated to a rare appearance by "The Rasterman". As we all know Carsten Haitzler or "The Rasterman", as he is better known, is the programmer who produced Enlightenment. His knowledge of X programming is extensive and he was able to show this when he did his presentation on performance programming for X-windows. He opened his talk with Enlightenment shown in all its glory on the screen in front of us. It does look so much better on a 70mm movie screen than on a laptop flat screen. Later on he went into the use of functions and other programming methods that are used in X-windows. Most of us thought it was brilliant. He was well received and much applause was in evidence after his talk. His other work includes Electric Eyes, Gtk theme engines, Imlib and several others. Carsten is a senior software engineer at VA Linux Systems in Sunnyvale California. He was originally from Sydney. He has also worked for Red Hat software.

By now it was getting late. Time to go home we thought? No, it was at this time that Alan Cox proceeded to give us chapter and verse on the latest developments in the 2.4 kernel which will be released soon. He asked us what it was we wanted to hear about and a show of hands produced an almost unanimous response that we wanted to hear about the 2.4 code, USB, PMCIA drivers and security fixes. It's all in there. He gave us his customary detailed insight into the latest kernel code. Alan presented his talk at the end of Friday and the night was setting in when he started. The audience weren't really interested in what the time of day was. They sat through to the end and loved every minute of it. Like watching a really good movie draw to a close.

At the end of the day we wandered along the road to bar 38 where about fifty of us sat down for a few drinks and a pleasant chat. A good night was had by all.

Saturday morning came around and Stephen Tweedie gave us a more than competent talk about clustering. If you've seen him in action you'll know what the word professional means. He works for Red Hat exclusively. It was reassuring to hear his well educated Scottish voice across the auditorium and the authority that he can show about file systems is more than slightly convincing. He's worked on DEC for a period of two years and also VMS kernels for high-availability clustered file systems. Great to see him in action once again.

After Stephen Tweedie, Stephen Lord stood up to give us some stuff about XFS. Stephen's presentation on the XFS journaling file system raised a great deal of interest from many people. One or two of them even woke up after the beer from the night before. Must have thought they were somewhere else? XFS would seem to raise more issues than you might at first think. So, if you do want to know more about XFS it is advisable to read something about it. Steve works for SGI. He is a senior file system developer. He's followed Linux development since 1993.

Michael Meeks came along after Stephen. He gave us an inspired talk about the GNOME component model. It was well received and more than a few laughs were raised. On his business card it just says "hacker". He's just a bit more than that. He's quite definitely one of the more intellectual and influential open source programmers around just now. Well worth going to see if you get the chance. We need more people like him. There aren't many around. He is presently in a 4th year at Cambridge University. He's studying Electrical and information Science.

Adrian Cox then started on the sort of thing that changes complete civilisations. Adrian's the man who brought us transputers. I think those of us who saw them and used them still miss them! In the present day he's working on a Beowulf in a box. Yes, that's a Beowulf in a box - not a pizza. His main problem he explained is getting hold of various types of hardware. He thinks he'll crack it quite soon. Most of his project is now finished. Just a few bits to tie up. After his talk he took the lid off his demo machine and a few of us had a close look at what was under the lid. This sort of thing shouldn't be missed.

After the long and slow death of Adrian's impressive presentation John Edwards came along from VM Ware. John is quite fluent as an advocate and demonstrator of VM Ware. In it's earlier incarnations there were problems. Since the 2.0 version came along things have changed quite a bit. John showed us that VM Ware can run NT4 and all the other windows systems very well under Linux. Even O/S2 and Linux can be run under Linux! As he explained, the people who test viruses do so with the help of VM Ware. The windows system is wiped out and VM Ware and the Linux system continues to run. In the question and answer session at the end of his talk John showed us just how much he really is in touch with all issues that are to do with software and hardware. If you haven't seen VM Ware why not try it now.

Next up was Owen leBlanc. Owen works at the Manchester Computing Centre. He is notorious for having written the Linux fdisk program and he explains that to everyone when he meets them. Owen explained to us that certain aspects of CODA are slightly unreliable but you have to be the system administrator to work that out. The security aspects are not that easy to grasp. However, the one outstanding feature of CODA is that it is completely reliable and very stable. His talk used several easy to understand pictures he is good at explaining things. The question and answer session was lively and had the effect of informing the audience in just about the right way about all things that are CODA.

Steve Whitehouse stood up next. Steve gave us a worthwhile talk about the global file system. He works for ChygGwyn Ltd. GFS is a journaled, fault-tolerant clustered file system. It gives high performance and a great amount of stability. His talk illustrated those facts very well. Many people showed great interest. Just now he's writing his thesis on "Error resilient image compression". He first became involved with Linux at the University of Swansea where he was an undergraduate. He's the maintainer for the DECNet kernel code. We wish him well for the future.

Heinz Mauelshagen gave us his presentation about a Logical Volume Manager for Linux. This is about a subsystem for online disk storage management. It's taken very seriously by the enterprise computing people and will help Linux to become more acceptable in future. An additional layer is added between the peripherals and the I/O interface in the kernel. Heinz did a wonderful job of explaining a complex idea to us. He is currently working on his 0.9 version of his logical volume manager. We look forward to more of it.

Wichert Akkerman gave us the future of package management next. Package type and management is always a controversial subject in Linux and the various types of packages that are available have led to more than a few arguments over what to do with them in future. The proponents of the Red Hat .rpm package think that nothing else exists. The .deb Debian package is seen by Debian users to be the only way forward. It is true that it's easier to install Debian packages. They give more information with simple and easy to understand error messages when something goes wrong. Wichert gave us a good account of these issues and then went on to discuss the prospect or possibility of a single package that could be used with Red Hat or Debian type systems. There was a debate at the end of his talk which was very much wide awake in contrast with some of the other brain dead question and answer sessions after some of the other talks. Wichert is an Msc computer science student at Leiden University. He works part-time for Cistron as a Linux developer. In January 1999 he succeeded Ian Jackson as Debian project leader.

Linda Walsh gave us Linux security policy at 6.15 pm. Time was rolling on but everyone was still very much awake. Linda left us in no doubt that we need a security policy for Linux. She went through her presentation at her usual pace, leaving no stone unturned. Security policy defines the allowed methods of access by processes to various objects in the system. Linda works for the Trust Technology group at SGI. She's also worked for Intel and Sun. She has an engineering degree in computer science from the University of Illinois. Not much chance of arguing with her!

After Linda finished her lecture we waited around for a short time so that everyone could gather together in a crowd. Once again we trooped off along the street to bar 38 at Hammersmith. Probably hasn't been that busy all year.

Sunday morning popped up over the horizon all to soon. There were those of us who didn't want to leave. We would rather discuss Linux. Hans Reiser started off with Reiser FS. Hans always gives the impression of a man who doesn't quite want to own up to anything. Don't let him fool you. He knows it all. His explanation of the Reiser file system was a masterpiece from beginning to end. Kind of thing that Leonardo Da Vinci would have been proud of. As Hans explained to us in his talk "this is just a very small step in the right direction". He thinks that everyone should be using systems that are fail-safe. Reiser FS is a journaling file system which uses classical balanced tree algorithms. Hans started Namesys in 1993 and employs twelve programmers. Sponsors for Reiser FS include SuSE, mp3.com as well as several other organisations.

Stephen Tweedie came back again to show us the EXT3 file system. He explained to us that EXT3 is basically EXT2 with a few bits added on. The audience showed interest and asked quite a few questions. There isn't too much information circulating about EXT3 just now. If you want to know more it's best to send e-mail to Stephen.

The final talk was by Luke Leighton. He gave us "Samba the next Generation". You might have heard the words "use the source Luke". Well, this is the real Luke Leighton. The one who works on Samba and MS Windows integration. The present round of Samba development is looking into integration with MS Windows 2000. By the account that Luke gave, it's probably best to sit back and wait for this one to come along. Luke isn't at all helped by the eccentricity of the MS Windows developers who as we all know don't have a clue about anything. Samba 3.0 development is said to be "well in hand". This probably means that someone remembered to pay the electricity bill! Luke works for Linuxcare, who also offer some very good training schemes. One of the features of Luke's talk was that there was a longish stand up exchange between developers about the thorny subject of how to integrate the ideas and concepts that were under discussion in the previous three days. At one point Hans Reiser was stood at the back of the lecture theatre with microphone in hand whilst having a discussion with Luke and several other developers. Where else could something happen in that way? Agreement over what to do with theses issues was quickly reached and we look forward to the next round of Linux development.

The future is Linux. Pick up a book on open source programming methods, read it, change the world.

Report on attendance at O'Reilly Open Source Software Convention, July 17-20 2000

(Sebastian Rahtz)


From the 17th to 20th July 2000, I attended the O'Reilly Open Source Software Convention (http://conferences.oreilly.com/oscon2000) in Monterey, California. I am very grateful to O'Reilly (through Josette Garcia, O'Reilly UK) for funding my entire trip, and the UK UNIX User Group for organizing the competition to send someone from the UK. It was a fascinating, if rather overwhelming, experience. As most of the talks are available on the web, I will not attempt any technical summary of what I attended, so please treat the following as a fairly informal personal view.

The conference was a huge affair, with some 2000 delegates gathering for two days of tutorials, a vendor exhibition, and two days of massively parallel talks. Folded into it was the 4th O'Reilly Perl conference, which was a big meeting in its own right. In between the formal talks, any number of parties and receptions, BOF sessions, and other social events were available to fit all the waking hours of the day. Holding it all together was a great deal of organisation by O'Reilly, and a lot of money; by which I mean not only e.g. sponsorship from Sun Microsystems, but also a general emphasis on business, and the way that open source software still makes money. No, this was not a niche meeting of slightly left-wing academic computer programmers (remember Richard Stallmann?), this was a demonstration of capitalist computing in the raw. Naturally, most delegates were from the USA, and the tone was resolutely upbeat.

What is this `Open Source'? In case, gentle reader, you have missed out on the politics of software development recently, things have moved on from the old dichotomy of commercial software vs free software, when we often confused the latter with `public domain'. Thanks in great part to the tireless advocacy in the 80s and 90s of Richard Stallman's Free Software Foundation, we now have considerably more precise legal software licenses. They vary in important ways, but the common factors are:

Associated, but not indivisibly, with this is the social programming movement popularized by Eric Raymond in his famous essay The Cathedral and the Bazaar ; this says that a loose community of programmers all contributing towards a software project can build something better and faster than a traditional hierarchical project. Somewhere on top of all this lies the general movement championed (hijacked?) by Tim O'Reilly and his company, Open Source . To some it means the Free Software Foundation ideal that all software should be distributed with source code, to others it mandates the Raymond methodology of `everyone equal' programming, and to many people it just means `good software for free'. To some it just means `anti-Microsoft'! All of these sentiments were visible at the Open Source convention.

The convention was organised around the theme of open source business strategies, and eight famous Open Source software projects:

  1. Apache, the widely-used web server
  2. BSD, one of the Unix-like operating systems
  3. Linux, the other...
  4. Mozilla, the open source of Netscape
  5. PHP, the embedded web scripting language
  6. Python, the scripting language that isn't Perl
  7. Sendmail, the humble slave lurking behind most mail delivery systems
  8. Perl, the ubiquitous do-anything language
The principal developers of most of these projects were on hand, and no doubt many important decisions were being made about them in quiet corners.


I opted to attend four tutorial half-day sessions:

I stuck with Python for the first day, with David Beazley doing a pretty good job of explaining the language, its philosophy, and its idiom. It came over as a pretty rational, comprehensive, language without the baggage of Perl's history. Indeed, at times, I almost felt inspired to go away and become a Python person, but David never quite pushed me over the edge. Perhaps the trivial thing that upset me was the use of indentation for delimiting blocks of program code. Perhaps if I had a computer to hand, and a copy of Python, I'd have forced myself, but as it is, I walked away still not a Python user.

On the Monday evening, there was a film show for those unable to face BOF meetings. I ate popcorn and watched trash.

One of the reasons I had been interested in Python was its associated with the Zope web application server, which was what I started to learn about on my second day. Unfortunately, Christopher Petrilli succeeded in putting me off Zope quite thoroughly, by never explaining what it was and and how it worked. I could not face the hard sell after a while, and escaped to part of Doug Tidwell's Working with Apache's XML Tools. Doug, an IBM XML evangelist, did a good job of teaching XSLT to his audience while I was there. While I did not learn anything new, it cheered me up to be in familiar territory.

On the second tutorial afternoon, I had been due to hear about Perl's XML::Parser module from Clark Cooper, but a glance at the notes showed that it was starting from a fairly low level with XML. Since the notes looked good for future reference, I then spent some more time in Doug Tidwell's session. Again, I almost came away convinced to use Cocoon, but not quite. As the conference proceeded, and I continued to compare more and more ways of serving up or storing or searching XML files, the less sure I became of which way to jump. Of which more later.

The Tuesday evening entertainment was a quiz show on Internet topics. 'Nuff said. I dropped in on a content management systems BOF, but it was degenerating into commercial product comparison, so I was delighted to find the film of Latitude on the TV.


The `real' conference started on Wednesday with a keynote address from Andy Herzfeld, a legendary figure from the team that built the Macintosh, now heading a new company building on top of Gnome. He was quite entertaining, but rehashing the history of Californian computing, and 20 year old stories about `Woz' (Steve Wozniak, Apple biblical figure) did not really do much for me, and Herzfeld did not really seem to have anything to say (although his company might very well do something important)

After the plenary, we had to choose between seven parallel tracks, which was hard. One could have gone to almost any of them, and learnt something. I, like many others, went to see Larry Wall deliver a sermon to the Perl conference. He took the opportunity to talk about music in an incoherent way, play a lot of instruments badly, and generally witter. No doubt the man is a genius of our times, but I could not take the tedious self-indulgence, and left. I expect the important information came after I went!

I could have gone on to hear Damian Conway explain how to write Perl programs in Latin (which really looked rather fun), but instead I wandered around the exhibition, with a variety of big and small companies and publishers showing their stuff. The noisiest was Jabber (http://www.jabber.com), who do instant messaging (based on XML). This looked like something that will suddenly get very big and is worth watching.

After lunch I went to a useful talk about the status of the various components of the Apache XML project; no major surprises, but the revised archiecture of the upcoming Cocoon 2 assuaged some of my doubts from the day before. In the new Cocoon, which is a Java servlet to deliver up XML on a web server, there will be a much more flexible scheme for delivering different content for different client needs, based on a configuration file (rather than the document, as at present). A scheme for rewriting links, depending on whether the client requested XML or HTML, was looking attractive, and I came away with the firm intention of experimenting with Cocoon 2.

For my final session of the Wednesday, I listened to Rael Dornfest explain the way O'Reilly's Meerkat (http://meerkat.oreilly.com) newswire service was set up, using PHP and XML. This was a good talk (the best speaker I heard at the convention), both entertaining and informative. Meerkat itself is a nice service. I came up with a second resolution, to experiment with PHP.

Possibly the less said about the Caribbean party in the evening, the better. Much more fun was the television premiere of Buena Vista Social Club , and you cannot ask for much better than that.

On Thursday morning, we had two keynote addresses, from Tim O'Reilly and Gregory Benford. O'Reilly (guess which publishing company he owns?) was a surprisingly poor speaker, both in delivery (he could barely put sentences together), and content (I think he was saying the the future is in `open services', information consolidation like their Meerkat), but his contribution to the politics of Open Source is so undeniable, I forgive him. Benford is a physicist and `hard' science-fiction writer. His talk about the digital culture of the future turned me off, I am afraid; when it became clear he was an old-fashioned American right-wing figure, I had to walk out.

My memory and notes refuse to reveal what I did on the rest of Thursday morning (I probably read email in the sumptious-setup computer room), but in the afternoon I spent some time in the session on graphics and Python, and got rather incensed with the PDF generator from ReportLab (http://www.reportlab.com). They seemed to be reinventing the DSSSL/XSL wheel, and making far too much unsound claims.

At this point in late afternoon I was emtionally and physically drained, and headed for the airport for the long flight home, and the joy of an unstarted copy of Harry Potter IV.


It is hard to sum up such a large affair. Inevitably, I only sampled a small proportion of what was on offer, and found it hard to identify a niche where I felt that I belonged. Despite the number of people attending, the range of topics was in some ways narrow, with the concentration on Apache, BSD, Linux, Mozilla, PHP, Python and Sendmail (and even then Sendmail and Mozilla did not really seem to figure much). Was this related to the book coverage by O'Reilly? The result of the breakup into specific software tracks was that the convention could be regarded as a series of introspective club meetings, with not enough interaction. It is hard, of course, to come up with an obviously better structure for the event.

My biggest surprise at the conference, perhaps, was the apparent lack of real discussion of the concept of Open Source, or much critical debate about eg copyright, licensing etc. Possibly this was all thrashed out in earlier OSCON meetings, but I had expected a little more navel gazing.

The organisation was exemplary, for such a big conference. O'Reilly clearly do these things well, and I applaud their commitment.

Table of Contents Next
Tel: 01763 273 475
Fax: 01763 273 255
Web: Webmaster
Queries: Ask Here
Join UKUUG Today!

UKUUG Secretariat