New Page 1

Open Source Development -- Background

What is open source programming?

Long history in "Brief History of Hackerdom".

Then Linux by Linus Torvalds in 1991 and on.

Then Eric Raymond tries to copy Torvalds methods in 1996? with fetchmail and writes about the result in CatB.

This essay gets wide readership and has made ESR the most famous spokesperson for open source development.

CatB inspired Netscape to release their browser source code for open source, as the Mozilla project. How has Mozilla done? (good XC topic)

ESR publishes book with the same title that includes other essays as well. (Homesteading is good XC topic.)

Topics from CatB

Read and discuss his point #1.

Read and discuss his point #6.

Read and discuss his point #7.

Read and discuss marked section, near point #8.

Read and discuss marked section, on page 46.

Read and discuss his point #13. (I actually agree with ESR on this.)

Read and discuss marked section on pages 57-58.

Read and discuss marked section on page 60.

Read and discuss marked section on pages 62-63.

Read and discuss marked section on page 70.

Topics from Dream On

What happened with fetchmail

ESR was a traditional manager.

He did all of the management operations that he says are not needed on open source projects.

Read and discuss each item.

Debugging... A (or the) major point of CatB... ESR did it all, based on suggestions from many testers. The debugging process was not really parallelized. The testing and sugggestion process was parallelized. But all bugs were fixed by one person -- ESR.

Lessons from Linux

Same thing. Linus acted as a highly centralized manager with complete control over the project. He makes all major decisions, including release dates, feature lists, who's code gets included.

Work slows down or stops when Linus is busy.

(Aside -- Linus has assigned major subsystems to trusted lieutenants, who have a lot of control over their area. But Linus makes the decisions about who is trusted and what they are working on.)

OSS (at least Linux and fetchmail) are not really leaderless babbling bazaars. They are, in fact, managed and built like the great cathedrals of Europe. There is one strong visionary/architect and lots of cheap labor.

What would a true bazaar-style programming project look like? Outline the steps from my paper.

(Since writing this article, I have learned that there are some small OSS projects that do run like this. There is a small team of highly-skilled trusted people who all act equally. Good XC topic. I can give you the examples.)

Finally, I speculate that successful OSS projects may be limited by the number of possible leaders who can act like Torvalds and ESR. The leader must be technically smart, have good people skills, and have lots of free time. It is well known that the number of good technical managers is small.