 | Level: Introductory Teodor Zlatanov (tzz@iglou.com), Programmer, Gold Software Systems
01 Nov 2001 The success or failure of any software programming group depends largely on its ability to work together as a team. From manager to members, to well-conceived, yet dynamic guidelines, the team as a whole is defined by the unison of its parts. Shattering the myth of the faultless programmer, Teodor dismantles the uninspired software group and then builds it up again into a synchronized, energized ensemble.
Welcome to a series of articles on developerWorks comprising a complete guide to better programming in Perl. In this first installment, Teodor introduces his book and looks at coding guidelines from a fresh perspective.
This is the book for the beginner to intermediate Perl programmer. But even an advanced Perl programmer can find the majority of the chapters exciting and relevant, from the tips of Part I to the project management tools presented in Part II to the Parse::RecDescentsource code analysis scripts in Part III. The words program and script are used interchangeably. In Perl, the two mean pretty much the same thing. A program can, indeed, be made up of many scripts, and a script can contain many programs, but for simplicity's sake, we will use the two terms with the understanding that one script file contains only one program. Goals of the book
Part I is full of tips to improve your Perl skills, ranging from best programming practices to code debugging. It does not teach you Perl programming. There are many books with that purpose, and they would be hard to surpass in clarity and completeness.
Part II will teach you how a small Perl software team can be better managed with the standard tools of software project management. Often, Perl programmers embody the "herd of cats" view of software teams. Part II will apply project management tools to a small (2- to 6-person) Perl development team, and will examine how managing such a team successfully is different from the classic project management approach.
Part III will develop tools to analyze source code (Perl and C examples will be developed) and to help you manage your team better. Analysis of source code is superficial at best today, ranging from the obvious and irrelevant "lines of code" metrics to function points (see Resources later in this article), which do not help in understanding the programmer's mindset. Understanding the programmer's mindset will be the goal of Part III. Tools will be developed that help track metrics such as comment legibility and consistency, repetitiveness of code, and code legibility. These metrics will be introduced as a part of a software project, not its goal. There is no perfection in programming, only its pursuit. Good programmers learn something new every day and continually improve their skills and technique. Rigidity and inflexibility are forever the enemy of ingenuity and creativity.
In pursuit of perfection
The most common mistake a programmer can make is not in the list of bugs for his program. It is not a function of the programmer's age or language of choice. It is, simply, the assumption that his abilities are complete and there is no room for improvement.
Arguably, such is human nature; but I would argue that human nature is always on the prowl for knowledge and improvement. Only hubris and the fear of being proven wrong hold us back. Resisting them both will not only make a better programmer, but a better person as well. The social interactions and the quality of the people, I believe, are what create successful software teams more than any other factors. Constant improvement in a programmer's skills and the ability to take criticism are fundamental requirements for members of a software team. These requirements should precede all others. Think back to the last time you changed your style. Was it the new algorithm you learned, or commenting style, or simply a different way of naming your variables? Whatever it was, it was only a step along the way, not the final change that made your code complete and perfect. A programmer shouldn't be required to follow precise code guidelines to the letter; nor should he improvise those guidelines to get the job done. Consider an orchestra -- neither static, soulless performers nor wildly improvisational virtuosos (though the latter is more acclaimed). A static performer simply follows the notes without putting effort and soul into the music; the virtuoso must restrain herself from errantly exploring new pieces of the melody or marching to the beat of her own drum.
Striking a concordant tone
Code guidelines are like the written directions a musician follows -- when to come on, when to come off, how fast to play, what beat, etc. The notes themselves, to extend the analogy somewhat precariously, are the goals of the project -- sometimes lone high notes, and sometimes a harmony of instruments. In an orchestra, there is a conductor that directs but does not tell every musician how to play, and everyone has a part in the performance. The conductor creates harmony. Because music has been around for many more centuries than the art of programming, perhaps these are lessons well worth learning. The software project manager is neither a gorilla nor a walled-off convict. She is a part of the team just like everyone else. The guidelines presented in this series are not to be blindly extracted into an official coding policy. The coding standards in your project are uniquely yours, and they reflect your very own orchestral composition. Don't force programmers to do things exactly right, thereby creating an atmosphere of distrust and fear. You can forget about code reviews, or admission of responsibility for the smallest bugs. Instead, present the guidelines and watch how people react. If no one adopts the comment format you like, perhaps it's a bad format. If people write without cleverness, perhaps you have been too clever in the guidelines. If the debugger you thought everyone must run is sitting in a dusty room, still packed, then rethink the need for Whizzo Debugger 3.4. Maybe everyone is happy with Acme Debugger 1.0 for a reason. Of course, programmers can be stubborn for no reason at all, only out of reluctance to change. It's hard to convince people that 20 years of experience do not entitle them to an organized religion. On the other hand, freshly minted college graduates often lack self-confidence. Recognize and adapt to those characteristics, and to all the others of your team. Present ideas to the stubbornly experienced in such a way that they feel they have helped with it. Build up the college graduates with guidance and support until they can fly on their own.
All this, just for a few coding guidelines?
Coding guidelines are fundamental to a software team, just as direction and harmony are to music. They create consistency and cohesiveness. New team members will feel welcome and gel more quickly. Ye olde team members will accept newcomers more readily. The loss of a team member will not cripple the project just because someone can't understand someone else's code. Keep in mind that speed is not the only measure of improvement in a program's code. Consider ease of testing, documentation, and maintenance just as important to any software project, especially for the long term. A language as flexible as Perl facilitates good coding in every stage of the software project. Although this book focuses on Perl, many of the principles are valid for other languages such as C, C++, Java, and Python. Finally, be an innovator. Regardless of your position in the team -- manager or member -- always look for new ideas and put them into action. Perfection may be impossible, but it's a worthy goal. Innovators are the true strength of a team and without them the melody grows stale very quickly. Stay in touch with your peers; continually learn new things from them. A medium such as Usenet (see Resources)
is a great place for an exchange of ideas. Teach and learn, to and from each other. Remember, there's always room for improvement. Above all, have fun, and let the music begin.
Resources
About the author | | | Teodor Zlatanov graduated with an M.S. in computer engineering from Boston University in 1999. He has worked as a programmer since 1992, using Perl, Java, C, and C++. His interests are in open source work on text parsing, 3-tier client-server database architectures, UNIX system administration, CORBA, and project management. Contact Teodor at
tzz@iglou.com. |
Rate this page
| |
