git历史2：bitKeeper做错了什么？What BitMover Got Wrong?

仍然是git历史，八卦文章第二弹

原文链接地址：http://better-scm.shlomifish.org/bk/what-bitmover-got-wrong.html

 

What BitMover Got Wrong?

Written by: Shlomi Fish
Date: 14-May-2005

Introduction

BitKeeper could have been big. It could have been used by almost all open-source projects that have migrated from CVS. It could have been the industry standard. It could have had the best developers and code from all of the other open-source version control systems. It could have sprang a prosperous community of companies and consultants. It could have been GPLed, while still not hurting the profitability of BitMover, BitKeeper's parent company. It could have received dozens of patches to the source code, weekly, some of them of very high quality.

But it's not. It is very pricey and its source is not available to the public. The gratis version has been discontinued, and the Linux kernel project, the crown Jewel of the BitKeeper's open-source projects, has decided not to use it any longer. (MySQL has decided to seek a suitable alternative as well, and so will probably most other projects). The entire history of the BitKeeper gratis version has been filled with flame-wars, license changes (often to the worse), and a lot of bad reputation to BitMover.

Was it worth it? Probably not. If BitKeeper had been a commercial-only product from the start, it could have saved itself of all of this frustration. But that's not the only path that they could have followed. This article will point what BitMover, and especially its vigorous (or some say flamboyant) leader Larry McVoy got wrong and how others can prevent that mistake.

What BitMover Got Wrong

BitMover got one main thing wrong and this is this:

It doesn't matter how many people abuse the spirit or the letter of the license, it matters how many of them would have been paying customers.

BitMover got it wrong all the time. Whenever, someone abused its license, they kept restricting it. Even the source was pulled out of the Net, because people have downloaded it and patched it to remove some of the built-in restrictions, whose removal was prevented by the license. Now, tell me - do people who have this disregard towards copyright, will ever consider buying a license?

On the other hand, this prevented many people from rolling their changes into the product. I recall that when I started to use BitKeeper there was something that bothered me and which I considered to fix, and wanted to send a patch. I couldn't because I did not have the source. On the other hand, I recently submitted some patches to some Perl modules on CPAN without even notifying the authors of my intentions, just because I had the source. I did the same with Subversion, The GIMP, and other programs, and most of my patches were integrated into the main product. We all know how successfulWikipedia is because it's world-editable.

BitMover has similarly employed several other restrictions in time. Yet, the original license and intentions were pretty acceptable to most open source users, who are not as uncompromising as Richard Stallman. Furthermore, once BitKeeper has gained a lot of momentum, it could have been converted to full GPL, without hurting the profitability of BitMover much due to the fact many people wouldn't deploy a version control system without the adequate support or endorsement of its parent company.

Unnecessary Paranoia

Judging by how Larry McVoy's is trying to protect BitKeeper, he thinks it should be a well-guarded castle, surrounded by a 1 km-wide chasm, filled with water with dangerous crocodiles, and with canons hurling Greek fire. So far, Mr. McVoy has:

1. Issued several patents on BitKeeper features. He said he plans to enforce them.
2. Posted a message to the Subversion mailing list where he warned them against integrating any of BitKeeper's features into Subversion.
3. Added the no-compete clause and is now trying to enforce it on paying customers as well.

All of these actions are pretty much useless and just leave a bad taste in the mouth and create antagonism towards BitMover. Even if you have patents, then trying to enforce them using a lawsuit against a well-funded competitor will drain the resources of BitMover, and create a lot of further antagonism towards it. Many small companies did not survive after they started a patent law-suit.

Even if people integrate some of BitKeeper's features and ideas into their own products, BitMover would better try to stay ahead of the curve and integrate newer features. That way people will always catch up with it.

Finally, if BitMover wishes to remain a company friendly to open source software, it must realize all of these things are a big no-no. Free Software advocates encourage competition. Those that respect the developers of proprietary software developers, believe that they should not try to prevent free software (and other developers of proprietary software) from duplicating their efforts independently.

A Final Note about Larry McVoy

To be honest, I still like Larry McVoy, BitMover's CEO and founder, and the main engineer of BitKeeper. I have sympathy for him, out of respect for him as an Engineer and as an Entrepreneur. However, I believe some of his beliefs and things he says frequently are completely false, and have no basis in reality.

Mr. McVoy, if you're reading this, please refer to the following resources regarding software management that you can read to get a better perspective:

• The Cathedral and the Bazaar series by Eric S. Raymond. Other writings by Raymond, who is widely considered as the foremost philosopher of the open source world, are also very good.
• The excellent Joel on Software site by Joel Spolsky.
• Paul Graham's Essays.
• David A. Wheeler's personal site.