Full Article Attached Independent Project Status Reports

Tuesday April 10th, 2001

Here are this week's reports. Featured are Beonex and Hermes. Enjoy!

#51 Re: Re: Re: go away.

by strauss

Monday April 16th, 2001 1:49 PM

You are replying to this message

I've never understood why people interpret the expression of concern as a statement that everyone should pack up and go home. When you have a problem, you solve it.

In Mozilla's case, the failure of the project so far results from some very comprehensible and addressable factors.

First, inadequate technical leadership has left the project in a stumbling, blundering state in which leads on key components can't even be bothered to send status reports. Get some seasoned management in there and get some discipline into the operation.

Second, short spending on quality assurance has resulted from adoption of false models of quality in the open source community, in which quality is magically invoked without cost by something called "eyeballs." Back on earth, here's what you need to do. Hire on a few dozen good quality engineers, get through that giant bug report backlog, and start looking at components using root cause analysis methods. Create test suites for all the major components and turn down any checkin that doesn't pass the tests.

Third, on the technical level, address the terrible bug-fix productivity among the paid contributors, few of whom seem to be able to fix more than two or three bugs a week. (See status reports.) If people can't fix a dozen or more bugs a week, you have either the wrong people or a system which is hard to debug (or both). In this case, we know that the incredibly long compile-link-run turnaround time on Mozilla is a major contributor to the slow bug fix rate. Find a way to boost the compile rates -- there are a number of ways to performance tune the compilation and linking process.

Fourth, get some competent user experience help. The one person on the project who seemed to understand this area left the project because no one cared about user interface standards and he was tired of being ignored. The UI for Mozilla is dreadful, rather like a Windows 3.0 application. No one who has even half a clue on these matters would approve an interface that is so inconsistent and incoherent. Hire a good UI consultancy and give them the XBL/XUL support they need to create an interface that is at least up to 1995 standards.

Yes, these things imply spending money. It's rarely possible to dig oneself out of a hole without buying a shovel. But if any more proof was needed that the free software/open source model does _not_ create great software for free, it's this project. Spend more now, or waste everything you've already spent. Your choice.