(Susan Small)
One of our members, Adrian Sherriff, has written to me to suggest that he would find it helpful to see a list of useful, interesting, and even entertaining internet addresses listed in the newsletter. Readers could suggest sites to visit, software to obtain etc which would hopefully be of service to the membership. To start the ball rolling, the following web pages relate to our forthcoming events and may give you some useful background information:
http://www.yahoo.com/Computers_and_Internet/Mobile_Computing
http://www.yahoo.com/Computers_and_Internet/Software/Protocols
Alternatively, if you're stuck for a witty quotation for
your .signature you could try:
http://pubweb.ucdavis.edu/Documents/Quotations/homepage.html
According to Carnegie Mellon University, there were 5,535,148 web pages available on the Net at the end of July 1995. Let me know of your favourites, those that you find really useful and if I publish "your" URL I'll send you a small gift, such as one of the rare, coveted UKUUG T-shirts!
(Mick Farmer)
At our recent meeting, your Council decided that it was a priority action to increase the visibility of the UKUUG, both as a way of showing existing members what services were available to them and as a way of attracting new members interested in our range of services. This decision provided us with a number of options for achieving this goal, all based on increasing the number of services available to members without increasing your subscriptions. We are now in a position to unveil some of these new services in order to gauge what you, our membership, think of these additional benefits of membership.
Historically, the UKUUG has always been known for the high technical level of its conferences and workshops and your Council wants to maintain that tradition. We are therefore planning an increase in the number of events that we offer each year. Here are the events planned for the next six months (more complete details of most of these events can be found elsewhere in this issue):
[email protected] (and also
[email protected]) can be either
The UKUUG is often asked whether we offer training courses and, at present, the answer is no. However, we realise that many of our members are consultants, educators, trainers, etc. willing and wanting to do just that. Therefore we intend to create a "Skills Register" of members. If you're interested, complete the details requested elsewhere and, within reason, we'll ensure that your information is available to callers at our Secretariat.
Of course, what's happening is a result of your Council getting together and deciding where we think the group should go. As always, we would love to hear your views on these services and any other aspects of the work done by the UKUUG. Don't deafen us by your silence!
I am especially pleased to include Richard Morin's article (The Joys of UNIX) in this issue. His mention of Jim Joyce reminded me of Jim at past EUUG (now EurOpen) conferences waxing elloquently about his early days with UNIX and his vision of the future of computing much of which has come to pass. He was always entertaining and good company to be with. As Richard says, he is widely missed by all his friends and acquaintences in the UNIX and Open Systems community.
I am pleased to bring you my first report following my election as Treasurer at the AGM last April.
To stand a reasonable chance of doing a good job over the next two to three years, I need to introduce effective controls on expenditure. This does not mean that no money will be spent, but that money will only be spent following consultation, and agreement on financial viability. However, it is important to remember that all historic decisions relating to expenditure will be honoured to ensure a smooth transition.
In summary, what I hope to achieve is that all costs (give or take a bit!) will be authorised prior to expenditure. Therefore it is essential that each "project" is supported by a budget submission.
In order to provide a consistent approach in the authorisation of "projects" and to ensure that "projects" will benefit the members, the following statements and plan will be used as a guide.
The relationship of the statements and plan can be considered as follows: Mission Statement – the reason why the UKUUG exists; Strategic Objectives – interprets the mission statement into objectives; and the Business Plan – how the UKUUG Council will meet these objectives.
I would expect that the Business Plan will change frequently to suit environmental (not the green variety!) issues.
Occasionally, organisations contact the UKUUG asking whether we run training courses in this subject or that subject. Although the group does not run such courses, we are willing to act as a clearing house to put organisations in touch with people willing to do the training. In this way an organisation may find someone local who is willing to do the work. To set this up, the group is creating a "skills register" of would-be trainers. If you are interested in being put on this register, please provide the following information.
1. Your name, contact address, phone number, fax number, and e-mail address if available.
2. Subjects in which you are skilled and willing to train in.
3. Your charging structure.
4. Geographic areas you are willing to cover.
The following information is also required before your details can appear on the register. This will enable us to ensure that only bona fide trainers are included. It will not be divulged to any third party.
5. Contact name and address of at least one person or organisation you have provided training for and who can verify your effectiveness.
Please send this information to the Secretariat at Owles Hall, or request further details.
I am delighted to be coming to the UK in October 1995 to present this tutorial. I will be in Edinburgh on 17 October and London on 19 October. The following is an Abstract of what will be covered:
The present of the Matrix in words, graphs, maps, and demonstrations:
The theme of the UKUUG Winter conference will be "Location Independent Computing".
With the increasing changes to the working environment, with ever greater connectivity, the bandwidth explosion and the push to outsourcing, work-at-home, distance learning, and remote management and support, the design of workable applications and provisioning their support infrastructure becomes an ever more skilled area. What are the considerations? Is latency solvable?
In many US states the law now requires a proportion of information workers to telecommute, a way of work that will become ever more relevant here. Technical aspects of telecommuting can be addressed in the topics above, the social and legal aspects are just as interesting. Will increased work at home regenerate communities? Will crime be reduced? What are the benefits to the companies who have to drastically change their approach to management?
With decentralisation, the "groupware" claims can now be tested. What experiences can you share?
All accepted authors will be expected to submit a paper in
electronic form conforming to the conference guidelines. Copies of
the guidelines are available from the UKUUG Secretariat, and
on-line at
http://www.ukuug.org/events/authors-guide.shtml.
You do not have to be a member of UKUUG to submit a paper. Submissions from speakers from outside of the UK are welcome.
[email protected], or may
contact a member of the programme committee directly.
Initial abstracts should be sent either by e-mail to
[email protected], or in hard copy format to the UKUUG
Secretariat. The committee would prefer all submissions to be in
electronic form. We will acknowledge all submissions.
Rik Farrow will be repeating the seminar he gave in February this year which was sold out. The venue has yet to be decided, but will probably be in London. Full details will be circulated as soon as possible.
For those who cannot make these dates, Rik will be repeating this seminar in Copenhagen the following week (see Calendar of Events).
Richard Stevens presented an extremely popular network programming tutorial during last year's IP Workshop at London Zoo. We have decided to ask Richard to repeat the tutorial in order to give a second chance of attending to those who were not able to be at the Zoo. The date has not yet been finalised, but will be in April or May 1996. Full details will follow as soon as they become available.
(Andrew Findlay)
Here's one for your diary! The October meeting will feature proponents of the various flavours of free UNIX explaining why their particular version is the best. Come and hear the champions of FreeBSD, Linux, and NetBSD. We also hope to have someone speaking on behalf of BSD/OS which, although not free, is reasonably priced. Come along and make up your own mind on what many people consider to be the true descendents of Bell Labs UNIX which are around today.
Andrew Findlay organises LUUG events, barndances, and ox-roasts. Between these momentous events, he is in charge of the team that deals with the exponential computing requirements of Brunel University.
(Martin Houston)
Mick has asked me to relay some news of happenings in the Linux SIG to the wider audience of News@UK readers.
I have been busy trying out the new Linux-FT release from Lasermoon myself. First impressions are very good. The copycache system really makes my disk space go a lot further without all the usual problems of running off CD .
Since its launch in January the group has managed to produce two newsletters maybe three by the time you get to read this. We are also seeing the stirrings of interest in forming local groups. The newsletter is always in need of material to fill it, and members to subscribe to it!
The last issue of Linux@UK -March/April was ready to roll in early May but the still quite small number of paid up members does not allow for separate print run to be made. At present Linux@UK print runs are piggy backed along with News@UK so we are pretty much a slave to the publishing cycle of the UKUUG's other newsletter.
The way to resolve this is… more members. The minimum print run below which your membership fees would be squandered in printers charges is a thousand copies. We can print some copies in anticipation of other people joining the group, but there is a limit to how much of the members' money can be speculated in this way.
The early days of the ACCU (or CUG(UK) when I was chairman) were beset by a similar problem. The first few issues of CVu had to be photocopied rather than printed and were working out too expensive to produce – certainly freebie copies of a 30 page photocopy could not be given away lightly! Luckily the Linux SIG have the UKUUG as a parent and our printing requirements can be merged onto theirs.
If you want the bi-monthly newsletters as promised then three things have to happen:
Firstly the ratio of mailing list to paid up group members is about 5:1 at the moment. What we need is a lot more Fence Sitters to make the twenty quid commitment to becoming official members of the group and getting their newsletter printed rather than just by the Web. Remember that if you are a full individual UKUUG member then joining the Linux SIG costs nothing – you just have to tell Owles Hall to put you on the list!
Secondly any Internet connected Linux users should spread the word about Linux and the SIG to your non net connected friends. The whole point about having a user group with a newsletter and local meetings is that it is not just a net thing anymore.
Thirdly we need material for the newsletter. News of local events. Second opinions of new (and old) Linux products that go beyond just reproducing product announcements. Do you know some really neat tricks with Perl? WRITE ABOUT IT! Think Tk is going to annilate Visual Basic? LET US KNOW! Do you have a good beginners question – preferably with an answer, others may find it valuable and time saving. Linux is such a huge and daily growing subject that nobody can hope to master all of it. We need to co-operate to stay on top of things and keep the forward momentum.
What perhaps has not been made clear in the past is that some of your membership money can be used to promote Linux – getting computer magazines to cover something so different from the Microsoft dominated PC main-stream is an up-hill task!
Press Releases, Brochures, Posters, attendance at computer shows are all things that we should be doing to increase the number of people that are given a fair chance to see if they want to join the Linux community. If they don't have enough interest to make the commitment then fine; but people should at least be made aware of the issues involved.
This all costs money – your money. However the more members we can get, the more publicity we can generate and the more members we can get. This is NOT a plea for charity. Linux has shown that in the field of software at least, co-operation with others is very much in your own self interest. Increasing the size of the Linux community is in the interests of every Linux user. Take one example – Linux device drivers for computer hardware. Already in some specialised areas, such as multi-port serial cards, hardware manufacturers are supplying and supporting Linux device drivers. As the global Linux community grows to several million strong AND is organised to have a lobbying voice then you would have a right to proper device driver support for ANY hardware you wanted to buy. You should also be able to go into your high street computer shop and buy a PC with Linux instead of DOS/Windows loaded onto it; after all it will cost the shop keeper nothing to do it. Support services for Linux – as with any complex software is a different matter entirely.
I would like to see a 10,000 member strong Linux User Group by the end of the year! That is enough to have a real effect on the structure of the UK computer scene.
The current explosion in Internet access had to start somewhere. Demon Internet started very small, its founders wanted to provide a low cost dial up service and got the founder subscribers to pay for a year up front to help finance the original equipment purchase. Just look at how fast the dial up Internet market is growing now! By joining the UKUUG Linux SIG you will be doing your bit to help Linux and Linux based skills become the mainstream of computer use.
Just by being able to say "We have a 10,000 strong user group for Linux in the UK" will be powerful persuasion for the commercial applications and support services to become available.
There is the potential for a whole Linux based economy spread over the country at a local level. If your car breaks down or just needs a service you don't usually have to refer back to the factory where it was made – a local garage should be able to do the job, and you can always shop around. There are little back yard garages and there are shiny dealerships with plush receptions and manufacturer accreditations. Both might do as good a job on your car – although as a general rule a big place that charges more will give a better feeling of assurance that the job has been done right. With Linux the situation for software should be the same – there will always be place for big software houses like Logica or Hoskyns that are needed to manage big projects. But there will also be more of a place for the small software house to thrive with locally generated business. Not forgetting of course the vast army of people who will just want to do it themselves with Linux because it is productive, cost effective and fun!
Why should software users have to put up with all the worlds woes concentrated into a single manufacturers support centre ? Great news for fans of telephone hold music but not so hot for the rest of us who just want to get the job done!
Linux with its freely available source code does not need centralised support. Skilled Linux users can sell their support skills to those with lesser or just different skills. It leads to a whole economy where software is demystified and skills are widely spread and widely available.
Software is now getting so complex that tearing down the old restrictive practices is the only real way forward. Programmers; don't worry – there is still an infinite amount left to do, and people willing to pay for it to be done!
The time has come to shout about Linux. To let people know that there is a real, practical and genuine alternative to the old ways of thinking that are leading inexorably to one giant Big Brother software company to control our lives.
Joining the Linux/UK SIG (if you are not already a member of the UKUUG) costs twenty pounds per annum. Membership runs for 1 calendar year from January to December.
Further details are available from the UKUUG Secretariat.
You may be aware that the United Kingdom is about to change the algorithm used to determine the changeover dates for British Summer Time, in order to harmonise the changeover dates within the EEC. Note that we will still use the GMT and BST timezones, so there is to be no common timezone in the EEC.
The change of algorithm means that many of our computers need a modification to ensure that clock time will continue to be displayed correctly. The algorithm used recently has been based on the Summer Time Act 1972, which defined that BST started at 02:00 GMT in the morning of the day after the third Saturday in March (unless that day was Easter Day, in which case BST started a week earlier), and that BST finished at 02:00 GMT in the morning of the day after the fourth Saturday in October. Parliament has in recent years reserved the right to alter the changeover dates to bring the changeover dates in the UK closer to those of other countries in the EEC.
The Seventh Directive (94/21/EC) of the European Parliament and of the Council of 30 May 1994 defined the start and end dates of summer time throughout the EEC, using a new algorithm with effect from 1996 (the algorithm being to start on the last Sunday in March, and end on the last Sunday in October, always at 01:00 GMT).
For 1995, the start date was 26 March: the end date will be 22 October in the UK and Eire, and 24 September in the rest of the EEC.
Note that effectively the rest of Europe has agreed to postpone the end of their Summer Time for a month, and in exchange we have slightly altered the changeover date rules to be the last Sunday in the appropriate months. Note also that the changeover time is now 1:00 GMT, i.e. an hour earlier than before.
We need to update computers with a new rule for 1995 (BST ends a week earlier than normal), and with a new rule for 1996 onwards.
TZ is set to a string
(usually GMT0BST), and the file
/usr/lib/tztab contains all the rules – the rules matching
the current value of the TZ variable are used. The
GMT0BST rules in this file currently are:
# Greenwich Mean Time, British Summer Time GMT0BST 0 3 25-31 3 1983-1984 0 BST-1 0 3 23-29 3 1985-1990 0 BST-1 0 3 25-31 3 1991-2038 0 BST-1 0 1 25-31 10 1983-1985 0 GMT0 0 1 23-29 10 1986-2038 GMT0(for the format of this file, see
man 4 tztab). This
table needs to be changed to:
# Greenwich Mean Time, British Summer Time GMT0BST 0 3 25-31 3 1983-1984 0 BST-1 0 3 23-29 3 1985-1990 0 BST-1 0 3 25-31 3 1991-1995 0 BST-1 0 2 25-31 3 1996-2038 0 BST-1 0 1 25-31 10 1983-1985 0 GMT0 0 1 23-29 10 1986-1994 0 GMT0 0 0 18-24 10 1995 0 GMT0 0 0 25-31 10 1996-2038 0 GMT0
/usr/lib/share/zoneinfo.
The machine should be set to use "GB-Eire" as a timezone,
and the file /usr/lib/share/zoneinf o/GB-Eire contains
the rules. Note that this is a binary file. The information in this
file is produced by the zic command
(/usr/etc/zic). The file
/usr/lib/share/zoneinfo/europe contains descriptions of
timezones in Europe. In the United Kingdom section, the section
under # Current rules currently reads (there should be
only two lines of rules):
# Current rules Rule GB-Eire 1981 max - Mar lastSun 1:00s 1:00 BST Rule GB-Eire 1972 max - Oct Sun>=23 1:00s 0 GMTThe starting rule can be left (which means that it was wrong before!), but the ending rule needs to be split into:
Rule GB-Eire 1995 only - Oct 22 1:00s 0 GMT Rule GB-Eire 1996 max - Oct lastSun 1:00s 0 GMTSo some movement of rules from 'Current' to 'Historic' needs to be done.
The work to generate the new binary file is actually performed
by the zic command which, according to the man page
(man 8 zic), takes the textual description file
(/usr/lib/share/zoneinfo/europe in the UK case) and
produces the binary files for each zone found in the textual
description file.
TZ is set to
GMT0BST, then programs which seek to derive UTC will also
use the USA algorithm (thanks to library code in various compilers
– I have the code in question for Borland C++ v3.1, for example).
Unfortunately one of the prime examples of such a program is the
RDATE client code used to synchronise the PC's
clock with a Time (RFC868) server.
The only practical way I have found is to set the timezone
manually to GMT0 or BST-1 twice a year,
AND reset the clock to the correct time. I have new code for
implementing a timezone variable which includes the rule (e.g.
GMT0BST-1,M3.5.0/01:00 ,M10.5.0/02:00). However this
cannot be used to set the TZ variable (since software
not written to cope with it would read that TZ was
GMT0BST and give the USA algorithm again).
RDATE client for the Macintosh
(called, I believe, "Network Time") which allowed the
rules to be set explicitly. However the version I found caused
problems this (1995) spring when it ignored its configuration and
changed late (on the first Sunday in April). My Macintosh users are
currently using K-Time (from Xinet's K-AShare
package, synchronising to the K-AShare server), and I wait to see
what happens this autumn.
Note that similar changes ought to be applied for the rest of the EEC.
Thanks to:
[email protected] ) for his
Network News article introducing me to the Seventh Directive
(94/21/EC) of the European Parliament.The APT Data Group is best known as publishers of newsletters such as Computergram and IT titles such as UNIX News. They recently launched an online version of UNIX News called UNIX News International and would like to alert members of the UKUUG to its existence.
UNIX News International is the online sister publication of UNIX
News and is aimed at managers who need to know about all aspects of
UNIX and open systems. UNIX News International appears monthly but
is editorially updated every two weeks. While it includes some
material common to the paper copy – namely the cover story, news
analysis and product reviews – it will also include a large amount
of new material and a grouping of resources that should make it the
most comprehensive source of UNIX and open systems available on the
Internet. Readers can access UNIX News International free of charge
by opening: http://www.globalnews.com/UNI/.
If you have any queries regarding UNIX News International, please e-mail or call me.
(Jane Morrison)
It has definitely been a very busy time since the last Newsletter was circulated.
Subsequent to the appointment of Ivan Gleeson as a new Council member at the May AGM, your Council actually attended a face to face meeting instead of the usual Tele-Conference.
Of course it's very difficult to come up with a meeting venue which suits everyone, considering that Mick, Andrew and Ivan are based in the South with Lindsay in Newcastle and Jim in Glasgow. Anyway the meeting was organised at the Stansted airport hotel on 21st June. It was a typical grey airport hotel but it served its purpose for the day long meeting which I think achieved a great deal.
If everything discussed and planned at the meeting actually works then it's going to be a busy time with at least three very good events between now and Christmas.
I am currently working on the December conference which will be held at the University of York. It is hoped that members will appreciate the fact that the venue is not in London and will easily be able to attend from all over the country. Coupled with the fact that the meeting will start at lunchtime one day and finish at lunchtime the next this should give plenty of time for travel. The package we are discussing with the University at the moment is fully inclusive and therefore the delegate price will include all meals and accommodation. This is something that many delegates have requested on returned questionnaires at other events.
At the time of writing, the Quarterman Tutorial at the impressive Bank of England looks almost organised and there is a possibility that it can be hosted again in Edinburgh two days before.
Event details will be sent to you once printed. Early booking is essential.
Your Council and the Secretariat all left the meeting at Stansted with a long list of actions. The next meeting – Tele-Conference will be held in the next couple of days when it will be interesting to see everything come together.
I have had one training day on the new SUN machine (all I could fit in due to being very busy). I am hoping to get more training over the next couple of months. It sometimes seems just as easy to get Andrew Macpherson to talk over any problems by phone as this usually solves things instantly… thanks Andrew. In the next couple of months OH will get an Internet connection which I'm really looking forward to.
Well, its certainly another very hot day here in Buntingford. We are in fact very lucky working at Owles Hall as it's not until late afternoon that the office actually gets too warm. Being a very old stone built building it takes quite a while for the heat to penetrate, but there again it take ages for it to get warm in the winter …….
This is not a half-hearted book. The authors (a journalist from Wired magazine, a researcher from Microsoft, and a senior scientist from Apple) do not like UNIX. Not even a little bit. They make that clear from the very first page: "Modern UNIX is a catastrophe". The rest of the book is devoted to proving this thesis. And they make no bones about their own fitness for the task "our judgments are keen, our sense of the possible sure, and our outrage is authentic".
Their view of UNIX can be summed up in the "UNIX Barf Bag" included with each book.
The book consists of the edited, annotated, and amplified highlights of the eight year life of the UNIX-HATERS mailing list. The authors take the postings from the list and use them as the skeleton of the book. This is perhaps the first, and major failing of the book. The posts are, to a large degree, written by people who have been pushed to the end of their wits by UNIX. Their posts are a kind of catharsis, purging them of their bile. As an alternative to taking an axe into the machine room, or tossing their workstation from the window, they probably have value. But in a book they have the air of a bar-room braggard about them. They are not rational, balanced academic discourses. Some are abusive, some vulgar, and some extremely funny. Some are all three. But they were posted to an uncritical audience – UNIX-HATERS membership is by invitation only. And invitation is offered only to those who have demonstrated their hatred for UNIX in a public forum. The intended audience is, I suspect, a super-set of this group.
Much of the content is informed by a vague nostalgia for the better times gone by – times when TOPS-20, Lisp Machines, Multics and home-brew operating systems ruled the earth. This nostalgia is coupled with a feeling that things could be done better.
The book is laid out in three main sections. These each set out to debunk a facet of the UNIX myth, as seen by the authors.
Part 1 (User Friendly?) deals with UNIX as seen by the naive user. In the first six pages of this section, UNIX is offered to us as: a virus, a plague, and a designer drug. In the second chapter we move on to guns and cars. The authors are fond of extended metaphor.
Beyond the metaphor, the authors target the inconsistency of UNIX commands. These are mostly well-known problems – the terse command names, the inconsistency of command-lines between similar programs, the problems of shell-expansion and poor error messages.
A few interesting points are made, beyond the usual complaints of obscurity. Why doesn't UNIX lock binaries that are being executed? Why is shell job control so complicated? And why is it not documented? No answers are forthcoming, apart from the odd cheap shot "UNIX doesn't have a philosophy: it has an attitude". No comparisons are made showing how much better other systems are. Is UNIX so much worse than all the rest?
An opportunity for a cheap shot is seldom passed up. The
eponymous mail program (mail in general merits its own
chapter) is used as an example of just how much "Unix
sucks". One section, which berates it for a particular bug,
ends "his bug seems to be fixed now. Or perhaps it just moved
to a different application". A few pages later, another
two-page bug description concludes: "this particular
feature…has been fixed". So why mention it at such length?
Are we to believe that only UNIX has ever been shipped with
bugs?
Discussion of mail is followed by a chapter devoted to UseNet. The authors seem to see UseNet as an exemplar of the UNIX culture, even though it is no part of the official operating system. This chapter, in particular, descends into simple name-calling, and plain insults. It is simply not about UNIX or the people who wrote UNIX. Laughing at the zoo of UseNet may be an amusing pastime, but blame for it cannot be laid at UNIX's feet. The authors even admit this "Usenet is not about technology, but about sociology".
The next target is terminal handling. Here the authors are on firmer ground, as this task is central to any operating system. And it is undeniably hard to set up a terminal for UNIX. And it seldom takes advantage of all full features of a terminal anyway. There is a fairly thorough critique of TERMCAP and CURSES (both written off as "half-implemented hacks"). No mention is made of the newer System V TERMINFO terminal database. Is it better? The same? Maybe it works, and just didn't offer scope for wisecracks. We aren't told.
From terminals we move onto window systems, and the X Window system, or rather "The X Windows disaster", in particular. The main targets of the contributors' ire are the complexity of setup, the size of the binaries, and lack of easy communication between programs.
The authors' view of X is summed up in the metaphor-salad verdict on the ICCM protocol, which is "a toxic waste dump of broken protocols … a twisted mass of scar tissue … to cover up the moral and intellectual depravity of the industry's standard naked emperor". In the case of X they do suggest an alternative, Sun's NeWS, which "solved the very problems that X was designed to create", whatever that means.
Part 2 (Programmer's System?) kicks off with a headlong tilt at UNIX shells. Here, the authors' dissatisfaction seems to be based around the fact that there are so many to choose from, "there is no standard shell in UNIX". Which ignores the fact that the Bourne shell is present on all UNIX systems. There is another set of gripes about inconsistent handling of symbolic links and NFS directories. The comments, or rather, war stories, from the mailing list are at their best here. The authors' comments add little.
Programming languages and environments are next on the hit-list.
Much is made of the paucity of tools available in UNIX – "most
PC debuggers put UNIX debuggers to shame; interpreters remain the
play toy of the very rich". The availability of similar tools,
such as SGI's cvd or CodeCenter, under UNIX is
dismissed as they don't come with the system.
There a few more gripes about well-known oddities in particular
programs, such as make and yacc. Many of
these are fair criticism, but are here presented as some kind of
conspiracy against the poor programmer. A quick, and accurate, tour
of the pitfalls of C is terminated with the implication that all
UNIX programs are riddled with bugs simply because they are written
in C.
C++ fares little better than C. Many people, even UNIX-lovers, would disagree with the authors that "C++ is fundamentally crippled". But what starts as a promising discussion of the language soon deteriorates into a criticism of programming style in C++ tutorial books. It then wanders off into a long discussion of the C pre-processor, which most C++ programmers would contend is unneeded in the language. The only really relevant attack is about the global namespace. And even that is too short.
Throughout this section no assement of UNIX as a system to write programs for is offered. Maybe UNIX is so bad that the authors don't want to encourage this practice. We aren't told what is wrong (or right?) with the system calls, standard libraries or memory management (apart from a paean of praise for garbage collection). The thrust is that UNIX has a poor programming environment, given the standard out-of-the box tools, when compared to, say 3rd party DOS tools, or a Lisp Machine. A fair comparison?
The final section (Sysadmin's Nightmare) is mainly a list of gripes about the difficult and non-portable nature of most system utilities, and file system tools in particular. Backup in particular gets given a hard time.
Despite the fact that few large UNIX sites use the
backup/restore pair as their primary archive method, these are
dissected in detail. And, unsurprisingly, found wanting. Other
targets of disgust are configuration files (why are there so
many?), the absolute power of root to do damage, the
dangers of SUID programs, and the general lax security inherent in
UNIX.
But the main ghouls in the sysadmin's nightmare are the file systems UFS ("the root of all evil") and NFS ("nightmare file system"). These two critiques contain most of the real technical meat of the entire book.
The section on UFS attacks many of the pillars of UNIX wisdom: files are untyped, there are no record-oriented files (both almost religious issues to those who have tried to feed program output back to a compiler on some operating systems) and lack of file locking. That robust file locking has been introduced in newer UNIXes is ignored.
Fundamentally, however, the attack is ill conceived, as fewer and fewer sites are using the original UFS code. Straw men are also raised: at one point, UFS is berated for being unable to cope with "intermittent hardware failures, SCSI termination problems and intermittent block transfers". The Berkeley FFS is briefly mentioned, in order to tar it with the same brush. Alternative file systems (AFS, Sprite, AIX's JFS) are briefly mentioned, in order to be dismissed out-of-hand as "nonstandard".
The chapter on NFS has each topic headed by a cute reinterpretation of the NFS acronym. Despite this whimsy, the well-known failings are discussed in accurate loving detail. These centre on unreliability, general lack of security and UNIX-dependencies. Interestingly, whenever the level of technical content rises, as it does in these sections, the number of posts from the UNIX-HATERS list goes down.
In the appendix a paper by Gabriel, of the MIT AI lab, is reprinted. This paper, though it is only 5 pages long, expresses far more coherently than the preceding 300, exactly what is wrong with (an admitted caricature of) the UNIX philosophy, and, unlike the rest of the book, Gabriel presents an alternative, as well as reasons why we must learn to live with UNIX.
Overall, this book contains many pieces of sound criticism, and is worth reading for these. It's just a shame that they have to be hidden amongst such a large amount of point-scoring, wise cracks, cheap shots, and "UNIX is a conspiracy" nonsense. It could be much improved if the posts from UNIX-HATERS were removed (or much reduced). It would also benefit if less space was given to humour, and more to presenting real alternatives to the problems discussed.
All in all, this book reminds me of Johnny in Rebel without a Cause "What are you rebelling against, Johnny?" "What have you got?". Despite the claim in the introduction that the authors and contributors are operating system writers, I suspect that this book will appeal more to dissatisfied users and undergraduates.
As Dennis Ritchie says in his Anti-Foreword: "it is not a tasty pie: it reeks too much of contempt and envy".
Suggested topics are:
I didn't get to UNIX early, but I have stuck around for a while. When Ken and Dennis were hacking B, I was learning to program on Stanford's experimental Wylbur system. Following that, I used about a dozen different operating systems, from single-tasking executives to plush time-sharing systems. My actual motivation for getting into UNIX was the fact that I had been spoiled by time-sharing systems like VMS. I was also entranced by the promise of graphics workstations, but couldn't afford the cost of a suitably equipped VMS system, ie a multi-tasking system, a reasonable set of tools, virtual memory, a bit-mapped screen, etc. I cast about a bit, floundered a lot, and ended up buying a Sun system. I liked the story the founders were telling, and I thought they might actually succeed, ie, delivering me a suitably equipped system at a price I could (barely) afford, and staying in business long enough to support it for a while.
Sun seems to have succeeded quite admirably, so I guess I made a good choice. More to the point, I could have done a lot worse; many of the vendors present at the 1982 USENIX trade show have long since gone to that great surplus store in the sky. Had I opted for a different UNIX vendor, my losses might not have been all that bad. I would have ended up throwing the computer away, but I would have kept the knowledge of how to use the operating system. As my original Sun (serial #285) has long since passed out of my life, I'm not sure how different this is from my current situation.
Had I bought a proprietary system, however, I would have had to learn yet another text editor, yet another command language, et tedious cetera. Having done this already several times, I could certainly have survived the experience. I've given up on volunteering for that kind of gratuitous pain, however, so I'm just as happy to have missed it.
The real loss, from the perspective of 1995, would have been the fact that I would have missed out on experiencing the joys (and pain 🙂 of UNIX.
Combining the new tools with existing UNIX features, I continually find new ways to think about problems. This is not a complaint! I like learning how to do new things, and even like finding out how to do old things in better ways. If I wanted a boring field, I wouldn't have picked Computer Science!
In short, UNIX provides me with a flexible and powerful set of tools, a stable and expanding software base, an interesting, adventurous, and cooperative programming community, and the opportunity to learn from others on a continuing basis. This seems almost perfect for a computer science dilettante like myself, and I would not hesitate to recommend UNIX to any interested programmer.
When a user can predict how a command will act, just by knowing how things should" work, s/he can save a lot of wasted time and effort. Consistency is important, and UNIX and MacOS both benefit from it.
Consequently, I think that we all need to proselytize a bit more. If you couldn't be as productive on a different system, tell your fellow programmers about it. Tell them, as well, about the flexibility and safety offered by the open systems approach. If you see a naive friend or relative getting ready to buy a PC, warn them about the deficiencies of that approach. Tell them about the pain they will go through in configuring new hardware and software. You may keep them from making a dreadful mistake. One major vendor reports that their multimedia CD-ROMs require ten times as much more support per user under Windows, as compared with MacOS. This isn't accidental, and some of this support load will fall on you, unless you're very quick. Why not avoid it entirely by steering them to a well-designed system?
Besides, isn't it a GOOD THING to keep the Microsoft Mafia from force-feeding the world with ugly, watered-down, inconsistent crud? Don't we want productive and elegant systems to survive and prosper? I think so, and I intend to keep making noise about it. Think of it as self-defense!
This first appeared in the July issue of SunExpert and is kindly reproduced by courtesy of the Editor.
In response to the growing demand for technical information by system and network administrators, the USENIX Association and SAGE, the System Administrators Guild today announced the program for LISA '95-its ninth Systems Administration Conference. The USENIX LISA Conference is widely recognized as the leading technical conference exclusively for and by system and network administrators from sites of all sizes and kinds. LISA '95 is expected to attract over 1200 professionals to the Monterey Conference Center on 18-22 September.
System and network administrators have some of the most demanding and critical jobs in computing. They are expected to keep increasingly complex networks, spread ever farther apart, running smoothly and securely. LISA '95 is designed to provide critical information on such emerging issues as integration of non-UNIX and proprietary systems and networking technologies, distributed information services, network voice and video teleconferencing, and managing very complex networks.
The keynote, given by John Mashey of Silicon Graphics, is entitled "Hardware, Wetware, and Software". Dr. Mashey will be bringing an INDY workstation and the software used in Jurassic Park. The two-day tutorial program will offer full and half-day tutorials, with expert instruction for administrators, novice through senior. Topics include: System and Network Security; Firewalls; DNS; Writing HTML and Web Server Management; Perl; Introductory System Administration; Solaris Administration; Sendmail 8.7; Communicating to Non-Technical Staff; and Performance Tuning. Other conference activities include Works-in-Progress sessions, an evening reception, evening Birds-of-a-Feather sessions and an informal table-top display featuring the latest tools from manufacturers.
A special workshop "Advanced Topics in System
Administration" will be held on 19 September. The workshop
will focus on the latest-breaking technical issues as introduced by
participants. There is no additional fee for the workshop, however,
participants must be registered for LISA '95. To attend, submit
a short proposal to the organizer, John Schimmel of Silicon
Graphics, via email to [email protected] no later than 1
August. The proposal should contain a topic for discussion and
explain why it is controversial. Workshop acceptances will be sent
by 14 August.
Since 1975 the USENIX Association has brought together the community of system administrators, engineers, scientists, and technicians working on the cutting edge of the computing world. The USENIX conferences have become the essential meeting grounds for the presentation and discussion of the most advanced information on the developments of all aspects of computing systems. SAGE, the System Administrators Guild, a special technical group within the Association, is dedicated to the recognition and advancement of system administration as a profession.
The 1996 USENIX Technical Conference Program Committee seeks original and innovative papers about the applications, architecture, implementation, and performance of modern computing systems. As at all USENIX conferences, papers that analyze problem areas and draw important conclusions from practical experience are especially welcome. Some particularly interesting hot application topics are:
Dates For Refereed Paper Submissions:
Extended Abstracts or Manuscripts Due: 18 July 1995
Notification to Authors: 31 August 1995
Camera-ready Papers Due: 14 November 1995
– System administration
– Systems and network security
– Distributed computing: DCE, DFS, RPC, CORBA
– Kernel internals: SVR4, Chorus, Windows NT
– Systems programming tools and program development
– Portability and interoperability
– Client-server application design and development
– Sendmail, DNS, and other networking issues
-GUI technologies and builders
An Invited Talks track complements the Refereed Paper track. These talks by invited experts provide introductory and advanced information about a variety of interesting topics, such as using standard UNIX tools, tackling system administration dificulties, or employing specialized applications. Submitted Notes from the Invited Talks are published and distributed free to conference technical sessions attendees. This track also includes panel presentations and selections from the best presentations offered at 1995 USENIX conferences and symposia.
The Invited Talks coordinators welcome suggestions for topics
and request proposals for particular Talks. In your proposal, state
the main focus, include a brief outline, and be sure to emphasize
why your topic is of general interest to our community. Please
submit via email to: [email protected].
[email protected].
Topics of interest include, but are not limited to:
Notification to authors: 30 July 1996
Revised papers due for shepherding: 19 August 1996
Camera-ready full papers due: 16 September 1996
[email protected]. For
administrative reasons (not blind reviewing), every submission (in
both its paper and electronic form) should include an additional
cover sheet containing: (i) paper title and authors, indicating any
who are full time students, and (ii) for the author who will act as
the contact to the program committee, his or her name, paper mail
address, daytime and evening phone numbers, e-mail address and fax
number, if available. The cover sheet mailed with the electronic
paper submission should be in ASCII to facilitate accurate on-line
bookkeeping and should be included in the same electronic mail
message as the PostScript file containing the paper. All
submissions will be acknowledged. If your submission is not
acknowledged by 21 May 1996, please contact the program chairs
promptly at [email protected].
Kovisoft, Inc. have announced that Universal Desktop version 1.0, a standard implementation of the Common Desktop Environment (CDE), is now shipping on HP-UX version 9.0.5, Novell's UnixWare 2, SunOS version 4.1.3, Solaris SPARC version 2.4 and Solaris x86 version 2.4. CDE is a joint effort of Hewlett-Packard Company, IBM Corporation, Novell, Inc. and Sunsoft, Inc. and is comprised of the best desktop technologies contributed by each company. Universal Desktop is a standard implementation of CDE originally licensed from Novell, Inc. which includes many software corrections and will be ported to all conceivable UNIX operating environments. Universal Desktop provides a consistent, feature-packed desktop Graphical User Interface (GUI) for users and Application Programming Interfaces (API) for software developers.
Ramana Kovi, President of Kovisoft, Inc., commented on the release, "We're proud to be the first vendor to ship pure CDE for several operating environments. Our plan is to provide CDE as the Universal Desktop with all the latest and greatest improvements and bug fixes in the most standard bundling possible, to many flavors of UNIX. It is our hope that our mix of standard packaging and good value to the end-user will help CDE earn its position as the de facto UNIX desktop."
Until October 31, 1995, the single workstation license of the Universal Desktop Personal Edition can be purchased for US $260. The Universal Desktop Personal Edition is priced at US $345 for a single workstation, US $695 for a workgroup bundle of three licenses and US $1,995 for the departmental bundle of ten (10) licenses. Each license bundle includes one set of desktop user manuals: CDE Installation Guide, CDE Administration Guide, CDE User's Guide, and CDE DTksh Manual. The Universal Desktop Developer Edition is available for US $595 and was designed specifically for software developers who plan to support the CDE standard. The Universal Desktop Developer Edition includes additional documentation such as the Application Builder User's Guide, CDE Style Guide and Checklist, ToolTalk Messaging Overview and the Internationalization Programmer's Guide.
The Universal Desktop includes features for software developers such as Electronic Mail API, Calendar Scheduling API, ToolTalk Development Libraries, Drag and Drop, Application Builder, Icon Editor, CDE Motif Libraries, CDE Motif Window Manager, CDE Widget Libraries.
The Universal Desktop includes tools for users such as Electronic Mail, Text Editor, Motif-based Terminal Tool, Login Manager, Session Manager, Workspace Manager, Application Manager (Program Manager), Calendar Manager, Print Manager, File Manager, Style Manager, Help Manager.
O'Reilly & Associates, is the only World Wide Web server available now for the millions of Windows users upgrading to Windows 95. WebSite allows Windows users to make their information available to any of the Web's estimated eight million users. The TCP/IP connectivity built into Windows 95 demonstrates the increasing impact of the Internet on current and future businesses.
"WebSite has been running on Windows 95 for months," said Gina Blaber, the WebSite product manager. "We've been shipping full-featured, Windows 95-ready software since May, and providing comprehensive tech support. Not only is WebSite winning in the testing labs, it is winning in the Windows 95 marketplace. To put it bluntly, WebSite is real and successful now, while other Web servers are still trying to get their first release to market."
"I was able to install WebSite in minutes and turn my Windows 95 desktop into a World Wide Web server," said Rob Wiese, network administrator at the University of California, Davis, Medical Center. "WebSite brings a powerful feature to Windows 95 users. My system now has the ability to serve information up on our internal system or, by connecting to the Internet, to make it available to anyone with a Web browser." "WebSite integrates smoothly and works great with Windows 95," said Chris Duke, Systems Engineer at Science Applications International Corporation (SAIC) and developer of WebBoard, a new conferencing system. "I was the first Windows 95 beta tester for WebSite and it has worked great ever since I received the first beta. I run the WebSite server on my office PC (486/33) and the server is so robust and efficient that I never notice it is getting the number of hits it receives. Its performance is fantastic, the tools are very intuitive, and I would recommend it to anyone interested in publishing information on the web."
WebSite offers several new Windows 95 user interface features, such as tabbed dialogs and tree displays, and Visual Basic support. Current data from spreadsheets and databases can be incorporated into Web documents using WebSite's special Windows CGI (Common Gateway Interface) along with Visual Basic, Delphi, or other Windows tools. WebSite's CGI can also accomodate ODBC and OLE automation within Web documents on your server.
WebSite works just as well on a LAN as it does on the Internet, acting as a powerful and easy-to-use workgroup server. A single Website installation can serve multiple workgroups through multi-homing and remote administration features. The Windows CGI support allows corporate users to integrate desktop tools like Excel and Acess. Web publishers can begin publishing with only SLIP or PPP account from their Internet providers. WebSite's graphical display manager for documents and links on the server allows an individual or administrator to run a complex Web site. With the access control features of WebSite, Webmasters can restrict access on all or part of their servers.
To set up a WebSite server, Windows 95 users need the following system requirements: a 386 or higher; VGA video display adapter; 3.5" disk drive; 12 MB RAM (16 MB recommended); 5 MB free disk space; TCP/IP connectivity; and, of course, Windows 95.
WebSite is created in cooperation with Bob Denny and Enterprise Integration Technologies (EIT). Bob Denny created Win httpd, the 16-bit server for Windows 3.1 which has been available on the Net and widely praised since early 1994. EIT, long-time developer of software tools for Web server administration, also founded TERISA Systems, the leader in Web security technology, and the CommerceNet Consortium, driving deployment of Internet electronic commerce.
SOFTBANK COMDEX Inc., the world's leading producer of Information Technology Tradeshows and Conferences, and The UniForum Association, the International Association for Open Systems Professionals, together announced that they have reached agreement to embark on an aggressive expansion plan of the UniForum Tradeshow and Conference Program, beginning in 1996. This multi-year agreement strengthens SCI's commitment to provide optimum venues for vendors and conference participants as they expand even further into the multi-platform environment of commercial, high performance computing. Further, the agreement helps fulfill UniForum's mission of providing state of the art educational programs to a growing audience of end users who need the skills and information necessary to implement open technology solutions throughout the enterprise computing industry.
Richard H. Jaross, Executive Director of The UniForum Association, said that "this agreement will bring together complementary strengths to create a powerful partnership that will benefit our two core constituencies: users and vendors of open systems. We've prided ourselves for years on providing the finest educational programs for UNIX and open systems professionals, and the logical extension of our mission is to bring this valuable service to thousands more through our new relationship with SCI."
This relationship calls for expansion in three different venues in 1996. Enterprise Computing Solutions will run concurrently with UniForum '96, on 12-16 February 1996 at the Moscone Center, San Francisco, CA. UniForum '96, along with Enterprise Computing Solutions, will both present comprehensive programs designed to address the needs of their particular audiences, including extensive conference programs, critical keynote addresses and key vendor participation on the exhibit floor.
These concurrent events, in essence marrying the enterprise computing solutions and applications development for business, with that of technical implementation and interoperability issues, provide a unique opportunity for all platforms to be presented within one environment and will have an impact in moving the industry forward.
A UniForum Showcase, comprised of exhibits and conference programs, will be an integral part of the Enterprise Computing Solutions '96/Windows World Show slated for 10-12 April at the Georgia World Congress Center, in Atlanta, GA.
Jason Chudnofsky, President and CEO, SCI, stated that "it was a natural migration for the two events to present a unified presence, as the issues and technologies of open systems and enterprise computing converge. Our timing is consistent with what we see occurring in the marketplace."
The UniForum presence grows again, with another UniForum Showcase at COMDEX/Canada '96, to be held in Toronto, 10-12 July 1996.
SCI first managed the UniForum Association-sponsored event last March in Dallas, Texas and future plans call for global expansion.
I've been giving the recent (ok, continuous) operating system wars some thought, and I think I have a new perspective on the problem. Perhaps instead of judging these systems on their technical merits, we should judge them by common social standards. Using an economic model, for example, leads to surprising results:
DOS is the most capitalistic operating system commonly available. If you are an application, you have free reign of the machine. CPU time (money) is available in any quantity that you wish. All resources are available for your use, and you (the application) bear responsibility for any problems resulting from your actions.
The socialist model is supported by Windows. You still are free in some sense to do as you wish, since the system is not pre-emptive, but the use of most major resources requires a system call, which will block and give other processes a chance to run. Clearly this is mandated sharing of common resources.
The loser in a social sense is clearly UNIX. It is the epitome of communism. CPU time is divided equally among processes, regardless of their own desire to excel. In fact, processes that use more than their fair share of CPU time are punished by having their priority lowered. Clearly this system fails to reward those (applications) who strive to be superior.
I'm going to remove all traces of windows and UNIX from my machine as soon as I finish typing this…
11-13 Sep, EurOpen Security Workshop, Stockholm, Sweden
18-22 Sep, LISA IX, Monterey, USA
18 Oct, The Matrix Today and the Internet Tomorrow, Edinburgh, UK
19 Oct, The Matrix Today and the Internet Tomorrow, London, UK
19-20 Dec, UKUUG Winter Conference, York, UK
15-17 Jan, EurOpen Security Workshop, Copenhagen, Denmark (DKUUG)
|
Tel: 01763 273 475 Fax: 01763 273 255 Web: Webmaster Queries: Ask Here |
Join UKUUG Today! |
UKUUG Secretariat PO BOX 37 Buntingford Herts SG9 9UQ |