Articles / Editorial: Ian Nandhra

Editorial: Ian Nandhra

Ian Nandhra, President/CEO of NC Laboratories Inc., sent in an editorial addressing the recently arised flamewars all over the net regarding the pretty weird situation of the LSB and other "Linux Standards" such as LSA, LCS, etc. He also discusses some of the basic Linux and Free Software concepts and he tries to show the difficulties a software vendor experiences that intends to develop/support a piece software for the Linux Operating System. Before anything else, I would like to thank Bruce Perens for his efforts on the LSB. Bruce's invaluable contribution has resulted in a great deal of discussion and the formation of other groups that will contribute in ways that we will see in weeks to come. The debates will continue.

Anyone reading slashdot.org this week will realise that it is almost impossible to say or do anything without offending someone on the Net. Slashdot's ( /. ) posting format (completely anonymous) makes this even easier. And complaining about something is much easier than doing something, and much more fun. So lets see how many people object to this:

Linux is about Choice,
Linux is about Freedom of Choice,
Linux promotes Tollerance and Community Spirit

Current rumor has it that a large vendor (lets call them XYZ) will select a Linux distribution to use. Other rumor says they will produce their own. Linux distribution "differentiation" results in a fragmentation which XYZ customers simply cannot tollerate. XYZ could choose one distribution, eliminating that problem and creating another. They are now tied to a single Linux vendor. The selection of one distribution means the automatic deselection of all the others. A risky strategy for XYZ who have just upset more distribution advocates than they have pleased. Perusing the reasoned debates on /. convince them further that this is not an entirely wise strategy.

The "problem" would not arise if there was one Linux - but source code availability means that there are as many potential distributions as there are people to make them. IE; lots. And if XYZ just secretly bundled an available distribution, news of their "choice" would get out into the community very quickly.

So back to basics. XYZ evaluates their baseline requirements;

  1. Scalability.
  2. Release Stability.
  3. Update Quality Assurance.

Their systems architects and engineers want a scalable system - too much fun to talk about here, lets get dull. The R&D Management want the lowest development overheads (something you never get with a fragmented technology) and low future maintenance costs (IBM maintains code spanning decades, current Linux code will probably still be in use in 20 years from now; source code availability is a very small part of this problem). The QA, Production and Sales people want a completely stable and transparent upgrade path between Linux updates. The Sales people want a seamless upgrade path for their customers (reducing their costs). The QA team want a way to test the Solutions Package to ensure that it meets their customers requirements. Testing involves specifications and specifications in turn involve Standards.

Standards are not the end of civilisation! They provide a metric by which a given set of code can be measured against functionality and portability. The POSIX.1 is amazingly successful. It has been incorporated and adopted into the most important portability specifications around (SVID3, 4.4BSD, X/Open's XPG4 Base and its successor, the Single UNIX Specification and thence UNIX98). It has been widely implemented on systems as diverse as VMS, MVS, and MPE/iX, as well as most traditional "UNIX" systems and more recently NT. Most of these implementations are either certified by NIST against the FIPS 151-2 test suite, or warranted by an X/Open UNIX Brand. Portability is about the ability to choose platforms suited to requirements (price, performance, networking, etc.) while still maintaining as much portability as possible for the application source-code developed to a particular specifications model. In the commercial market, it does not matter if the platform is UNIX, Linux, NT, VMS, MVS. It's whether it supports the relevant standards specifications (X/Open, POSIX, NIST FIPS) required by the application that matters. A long time friend used to constantly remind me that: "It's the Applications, Stupid" It was both a joke and a reminder that it's the APPLICATIONS that matter in Linux. And applications porting in the commercial sector require standards conformance.

Linux was certified against FIPS 151-2 in 1996 and these Certification tests (and others) are being re-run to see how far Linux has drifted from this original certification (anyone interested should drop me an email).

The POSIX.1 standards tests in question are, ironically, freely available. Yet the mere mention of them causes unbelievable amounts of hostility from people that should know better. The mention of standards bodies, or worse still, regulatory boards - just don't even think about it. Existing standards are usually ignored, bypassed or modified as a convenience measure by those with something to loose by their adoption and the consequences to those who rely on these standards is simply ignored. These existnig standards endure because developers and customers value them.

Linux distribution developers must take into consideration that their actions can have huge consequences for those building products on Linux. The argument that Open Source cures this problem (ie we can just type make) is no use if there are tens of thousands of installations involved across a few hundred suppliers and organizations. Freedom of expression carries with it the heavy responsibility to be considerate of others.

XYZ have established test and QA proceedures for existing platforms, and a standards conformant Linux would fit right into an existing model. But this would result in XYZ Linux, which might not be wise and does not fit into XYZ business model. However, if other ISV's with similar problems (and very large numbers do) grouped together a single commercial Linux would be possible, aimed at solutions providers. Lets call it "Enterprise Linux" for now.

Linux is supposed to be about Choice. Right?
Linux is supposed to be about Freedom. Right?
Linux is supposed to be about Expression. Right?

So there should be nothing wrong with Enterprise Linux. Right?
Especially as only those that want to use it will. Right?

Well, yes, and no. The potential losers would be the existing Linux Vendors who would (a) have a competing Linux distribution and (b) loose market share. Even worse, there would be limitations on the ability to retain users ("Brand Loyalty").

But the Open Source model ensures that the source code to Enterprise Linux will always be available and those Linux Vendors wanting to can use it as they wish. Or not. As they choose. It's about choice. Right?

If Linux really is about Freedom of Choice, organisations like XYZ or anyone else introducing new concepts, be it Standards or Web-Radio, will be welcomed for the contribution that they make. Real World Economics dictate that Software will be both sold and freely available for many years to come. The users have a choice - they dont have to use it if they dont want to. They also have the right to choose without the risk of being flamed to death on forums such as /. and others just because they choose one thing over another.

I believe that Freedom of Expression and Choice is a Right. We enrich our own lives and others around us by mixing these choices. But this is no Free Lunch and this right carries a responsibility with it. The responsibilty to extend the same right to others.

The flame wars and intimidation about issues such as Gnome/KDE, LCS/LSA/LSB, RPM/.deb, source/binary distribution, right/wrong, good/evil ;-) insult the very goals we hold dear. As Freedom of Expression gets shouted down, freedom of choice withers and dies. And we all loose. It is not a crime to use commercial software, nor is it a crime to use a Microsoft product. It is a crime to abuse those that choose differently from you. It is a crime to publicly and privately attack them for being different. Showing them a cool graphic with GIMP is more likely to convince them than a public flaming about Photoshop on Win95. Chances are they will install Linux and use Gimp ;-) Or is ignorance a crime as well?

It's about freedom and choice, right? The next time standards and their organisations, Enterprise Linux and alike come up, pause a little and wonder what this will contribute and why those involved are trying to climb the mountains. There is precious little money to be made and no thanks to be gained. The organisations and people behind them are contributing to Open Source and to the applications developers that rely on Linux. Just because something is new does not necessarily mean it is bad. What would have happened in 1991 if a Finnish student was ridiculed for announcing 0.02 of Linux? Then as now, it is hard to predict the concequences of our actions.

This editorial has strayed away from the LSB theme. But the events of the last 2-3 weeks strayed very, very far away from the "Linux Way" and the ideals I for one thought we cherished and defended. It is possible to contribute to Linux without spending 22 hours a day posting to the Net or writing emails. Lack of high profile lime-light exposure could indicate shyness or a small ego rather than conspiracy theory plot to kidnap a penguin :-) Lack of exposure does not mean lack of value.

Have a great weekend, and as a comedian in the UK used to say,

"May your God go with you...." (unless you use vi) ;-)

Ian Nandhra

Screenshot

Project Spotlight

Kigo Video Converter Ultimate for Mac

A tool for converting and editing videos.

Screenshot

Project Spotlight

Kid3

An efficient tagger for MP3, Ogg/Vorbis, and FLAC files.