Archive of UserLand's first discussion group, started October 5, 1998.

Review of Mozilla M4

Author:Eric Kidd
Posted:4/24/1999; 11:15:47 AM
Topic:jwz resigns
Msg #:5279 (In response to 4744)
Prev/Next:5278 / 5280

Not too long ago, I asked:

Did jwz quit too soon?

Jamie Zawinsky, "head curmudgeon" at Mozilla.org, resigned just before the first anniversary party and fired off a scathing article. I agreed with many of his criticisms and added a few of my own. Mozilla was uncompilable and it couldn't browse the web. The license was broken. The code was a mess.

(People who aren't familiar with Mozilla might think that Netscape released the code for Navigator 4.0 or 4.5. Nope. Instead, they released the code for 5.0, which didn't actually work yet.)

The Dirty Secret

Navigator 5 was a doomed product. Netscape wrote Navigator 0.9 from scratch in less than a year and never rewrote it. They moved from version to version, piling on features. Nobody ever said, "Stop! Lots of things have changed since 1.0. We need to redesign stuff!"

Things were starting to break down by Navigator 3.0. Netscape needed to step back, reassess their product, and take the time for a major overhaul. But they were under attack by Microsoft and didn't dare to skip a browser generation.

Navigator 5 was still based on the original codebase. It was full of #ifdefs, Win16 support, fake multithreading hacks and layers of useless code.

Biting the Bullet

At the end of last summer, Mozilla killed the project that would have been Netscape 5. They realized that world didn't need another fat, broken, slow browser that didn't support web standards.

Instead, chose to skip straight to the "next generation"--the browser that they really wanted to be building. This decision cost them lots of time, and probably set back their eventual ship date by close to a year.

The new browser would have a brand new layout engine with support for CSS and XML. It would be more modular, more scriptable and designed for future growth. The user interface would be portable--each dialog box would only need to be implemented once, instead of once per platform.

The Risk

I've only been programming for about twelve years, and even I get suspicious when somebody tells me this story. It's a very old story. "We'll build the biggest, greatest and best program the world has ever seen! Finally, we'll do it right! When we finally ship this sucker, it's gonna be the best of the best."

Anybody remember OpenDoc? Taligent? The original promises for Windows NT 5? The "best of the best" is usually very late, and too slow to be useful.

The Current Status

The Mozilla team has a series of milestones. (You can read about them on the SeaMonkey page.) The first usable version of Mozilla, at least on Linux, occured at the M3 milestone. Right now, I'm running the M4 binaries.

The layout engine is gorgeous. It still has some bugs, but they're normal bugs for a pre-alpha release. Pages display quickly (even Slashdot), and all sorts of new CSS features allow me to write impressive pages which degrade gracefully in older browsers.

The portable UI isn't quite so far along. It's beginning to work, but still has problems with "platform parity" (which is just a fancy way of saying it works better on Windows than elsewhere).

The network code is still broken. Netscape recently announced the NetLib project to overhaul the old, single-threaded code and replace it with fast, multi-threaded code.

The Good News

Mozilla's hitting the milestones on time. Their development process is well-managed, publicly accountable and moving quite quickly. Netscape's still the single biggest contributor of code, but the module owners are seeing lots of outside bug fixes, third-party ports and useful features. An outsider wrote the Mozilla ActiveX control which lets web developers run Mozilla as an IE plugin. (Sneaky and devious, isn't it?)

For the first time in a year, Mozilla "feels right". I've been using open source software for a few years now, and I've seen lots of projects go through this critical stage. The basic architecture feels right, the core of the product works, and each new release is much better than the previous one.

Most open source projects which make it this far succeed, and eventually deliver stable, working software.

The Prognosis

About nine months ago, somebody asked me what I thought about Mozilla. My answer: "I'll believe in Mozilla when I can actually use it to browse the web."

Today, I can actually use Mozilla to browse the web. It doesn't work very well yet, but it's improving quickly.

Cheers, Eric

There are responses to this message:

This page was archived on 6/13/2001; 4:49:31 PM.

© Copyright 1998-2001 UserLand Software, Inc.