Netscape code dropgreenspun.com : LUSENET : Joel on Software : One Thread
On Joels comments on Netscape he appeared to miss the point about the Mozilla project: the old Netscape code -was- unservicable. It literally could not be patched anymore and the architectural problems meant simply moving things around was not practical. They had three different engineering teams for the three main platforms, and they effectively had three versions of the code for each platform. It was a mess.
So they started over. Stupid maybe, brave certainly. Mozilla has taken such a long time because they were committed to crossplatform development. If they had decided to go for a Windows only browser (which they could have done), they wouldn't have needed to reimplement COM into XPCOM, or use the Netscape Portable Runtime, or design XUL for user interfaces. They could have just reorganised the rendering engine, and shipped. The result would not have been even 5% as powerful as even the 0.9.3 version of Mozilla is. Although as a commercial browser project Mozilla would have been judged a disaster, as an open source project to develop a browser and also a robust cross platform development system it is a resounding success. And they have taken all of Joels Test points to heart: they don't just use daily builds, they use constantly repeating 24 hour builds. They don't just have a bug database, they have an open web based bug database (which they created themselves as there were no web based bug databases which met their needs). They are determined to fix as many bugs as possible for 1.0 and the result will be a program that is smaller, faster, less buggy and more powerful than Internet Explorer is after years of development.
IE is a great example of when you -should- start again from scratch. IE's support for XML has been a total hack, it turns it into HTML internally as the rendering engine can't cope with proper XML like Gecko can. This will not only limit IE but due to it's market share hold back the web too - IE doesn't support MathML, SVG and ChemML like Mozilla simply because it cannot. The web and its technology has changed too much for just a few simple macros and reorganisations. While IE lags behind, Mozilla will be getting better every day.
Thankyou for otherwise excellent and informative essays -mike
-- Anonymous, September 13, 2001
don't bash ie that much. i've downloaded the latest build of mozilla a few weeks ago, and tested their html test pages. this went along nicely.
then i put ie6 on the left side of my screen, mozilla on the right side. i then drag n dropped mozilla test pages in each browser. ie was always faster, sometimes so much it amazed me (granted, i only have a poor p2 350).
xul ? i like the concept (well i love xml so i'm biased ;)), but why didn't they use it to create win32 controls on windows, xxx controls on xxx platform ? if someone knows, please tell me. i can't say i like those custom controls very much.
svg ? there's the adobe player, which has been updated a few days ago. tested it just a little bit : works fine.
i'm not a microsoft fan, nor a mozilla fan. i merely see what works best for what i do - and right now, it's ie.
-- Anonymous, October 04, 2001
thinking of it, ie lags behind for PNG support. THIS is truly a shame.
-- Anonymous, October 04, 2001
There is no "point of no return" where it gets impossible to add new features or fix bugs. Every time you make a change to the codebase, you have a choice. You can either make it better, or you can make it worse. If you insist on making it a little bit better every time you check in a change, it is absolutely possible to turn things around. Without this attitude, even a complete rewrite will degrade in quality before it's even reached status quo with the old, "unmaintainable" system.
-- Anonymous, October 05, 2001