Deliver Us From Linux

As a Microsoft developer (I am not an employee, nor am I a "certified" developer), I am glad someone finally spoke out with regard to this very topic ["Can Linux survive its unreasonable defenders?," December 1999, p. 6]. If I hear that Linux is the answer to the meaning of life and the "messiah" that is going to deliver us from Microsoft one more time, I think I'll be sick.

I am by no means an avid Microsoft supporter. In fact, I consistently find limitations in their tools and OSes. Furthermore, Microsoft (and Mr. Torvalds for that matter) has only provided us with the tools to do our jobs. I don't care if it's a winner-take-all industry or if it gets "shared" and I have numerous choices. It's only a job and these Linux jihad-types need to let it go.

Nothing is perfect, and Linux in many ways fits this mold as well ... everything can be improved. I have heard it all over the years; I believe the last operating system that was hyped as the "end all, take over the world OS" was something called OS/2. And while it may have had some advantages, it wasn't what it was hyped up to be. I didn't have enough fingers and toes to count the times our OS/2 network was down on a weekly and sometimes daily basis.

So far, Microsoft has provided me with all the tools and database technologies to do my job, as well as a fairly stable OS environment to run them in. I see no reason to switch. While Microsoft might not serve every purpose as well as other tools or OSes, they serve my purpose, and that is all I care about. If they die someday, then I will learn the next tool and continue to make a living — it's that simple.

It's a shame that the much-revered "open" source code people are so closed to criticism and comments. Personally, I think it's just sour grapes. Microsoft will listen quietly if you call to mention a bug in one of its products, and will even sometimes solve it for you. It's an imperfect industry full of imperfect products. If the people that so rudely blasted Mr. Williams would take the time to offer some much needed support for Linux, they would, as you said, be doing much more to further their cause.

Chris Hylton

The Best of Intentions

Regrettably, the article by John Williams ["Is Linux ready for the enterprise?," October 1999, p. 92] was not one that I read, but upon seeing Mr. Bucken's editorial ["Can Linux survive its unreasonable defenders?," December 1999, p. 6], I did read the letters to the editor. They brought an old saying to mind: "People living in glass houses should not throw stones."

I've been around computers and databases for the last 15 years, and any software install can be botched by the most trained and best-intentioned individuals. Just because one person had a hard time is not an opportunity for the rest of us to gloat.

On the Linux front, I've installed it at home along with a trial copy of Oracle 8 for Linux. The Linux I used was RedHat 5.2 and, in the end, I ended up installing it three times before I got it right. To Red Hat's credit I always got a functional system when I was done. Getting it right mandated a little RTFM on my part, which resulted in the second and third installs ... when I finally did get it right.

Are there still items that need fixing? Sure, but so does HP, Microsoft and so on. There still is no good substitute for a thorough understanding of the software, installation process and requirements of the final product. In Mr. Williams' case, it would appear from the letters to the editor that he did what any very competent IT person would do, find a test box and experiment.

Good show. As for Mr. Williams' detractors ... your day is coming!

Richard Goulet

Got a Problem? Read a Book!

I enjoy Application Development Trends and find it a useful and timely source for what's happening throughout the industry. There are always articles and comments that apply either to my work or my broader interests.

I read John D. Williams' October 1999 column, "Is Linux ready for the enterprise?," [p. 92] with keen interest as I was in the middle of a similar project. After reading Mr. Bucken's editorial ["Can Linux survive its unreasonable defenders?," December 1999, p. 6] and the Letters column, I wanted to throw my two cents in.

I have five networked desktop PCs at home. One has Microsoft Windows NT 4.0 loaded as a standalone server (non-PDC) and the others have MS Windows 95. To this, I wanted to add a Linux box so that I could explore the world of Linux and complete the final project for my Masters in Computer Science — to integrate Windows NT and Linux on one network.

My first choice of commercial Linux packages was, of course, RedHat. Big mistake! Nothing went smoothly and nothing worked. If Mr. Williams thinks that Red Hat's customer service people "replied in a very nasty tone" after only one inquiry for support, he should have been in my position of having to get Linux up and running in a short period of time. After several calls over a few days they became downright obnoxious! We, Red Hat and myself, parted ways with Red Hat maintaining I was trying to install Linux on unsupported hardware. Red Hat is definitely off my Christmas list ... permanently.

While browsing the shelves at a local computer store for an alternative, I came across a copy of Linux Universe, 3rd Edition, by S. Strobel, R. Maurer and S. Middendorf [New York: Springer, 1997]. The description of the Linux distribution included with the book was impressive and contained everything I was looking for. I brought a copy home and after reading the installation chapter, had as smooth an installation as any I have experienced with the MS Windows OS and applications. In a very short time, I was greeted with a login prompt, and, as they say, the rest is history.

The Linux box has been up and running on my network now for several weeks and I have experienced no "hardware incompatibility" problems as Red Hat maintained. The only bump I experienced was getting Samba configured correctly, but that was the purpose of my project — to learn to integrate NT and Linux. The Linux box acts as a WINS server for my network, and I am able to share files and printers between all of the networked PCs, regardless of OS. I have recompiled the kernel to trim it to my hardware, and also downloaded and compiled the latest version of Samba. Both of these tasks went off without a hitch. I would not hesitate to recommend the Linux Universe distribution to anyone, experienced or novice.

John M. Farahay
[email protected]

I Was Just Thinking

As always, the recent issue of ADT was an interesting and informative magazine that I thoroughly enjoyed. However, the following comes to mind. The information in the sidebar "CMM: Five levels of maturity," [January 2000, p. 36] matches what is in the textbooks; logically, however, the labels on Steps 2 and 3 should be reversed. Level 2 should be defined but lacking the methods to make it repeatable, while Level 3 should be a refinement that allows one to repeat the methods of Level 2 by managing them through self-discipline or maturity. I never could grasp how something could be repeated before it was defined.

This may be semantics, but it shows how the use of a single word or phrase can confuse those who have not thoroughly studied the topic.

Similarly, Level 4 (Managed) would be better understood as Measured so that it is not only repeatable but also predictable. Level 5 would then use data from Level 4 to facilitate improvement presumably through benchmarking, otherwise the processes would have been instantiated differently to start with if one could deduce the best ones independently.

One question that I have asked at CMM classes is whether achieving CMM Level 3, 4 or 5 improves software quality and other organizational attributes, or whether those firms that are successful happen to display characteristics of a CMM. An interesting follow-up article might cover such issues.

The main article ["The changing rules of CMM," January 2000, p. 35] made an excellent point about shifting away from CMM as religion. One size fits nobody. Until there is one standard organization structure and meta process, any CMM, QA, SEI or assorted methodology/process will need to be tailored in order to gain the maximum benefit for the actual, existing organization. It is regrettable that the DoD is requiring CMM as a prerequisite for ACAT 1. They are essentially saying that the government does not know how to measure results, so it will instead micromanage the methods.

I would also like to see articles on CMMI and various systems engineering and related higher level initiatives, such as EIA 731 and ISO 15288 and 12207. Real productivity gains cannot come from an isolated emphasis on software or "quality" without considering the entire enterprise as a meta system and ensuring that supported, relating, dependent and concomitant processes all work together for the optimization of the entire enterprise. Far too many people either ignore or are ignorant of the fundamental law of systems and charge forth with faulty assumptions as to what will work best.

Microsoft has reportedly rejected the idea of using CMM, saying that they are better. They are certainly fast to market; the assessment of the quality of their products is left to readers to decide for themselves.

While a widely accepted viewpoint although not a standard, I must disagree with the definitions in "ASQ is more than just testing tools," [January 2000, p. 55]. The article says a process is higher level and more general than a methodology, which is lower level and more detailed. A framework is a generic architecture I would liken to an Ada package, and has nothing to do with methodologies and processes unless such are included in it as would be done for a very high-level or organizational view. Having suffered through some 40 methodologies, I can attest that they were usually imposed using a religious silver bullet fervor not unlike some of the newer processes.

Finally, the article "Containers: A sign components are growing up," [January 2000, p. 41] cleared up in an hour what a two-day class left me puzzling over.

William Adams
[email protected]