Weekend Discussion

Friday March 3rd, 2000

This weekend's discussion: What keeps you interested in Mozilla, and what expectations do you have?

Just click the responses link below to enter the forum. Let us know what you think!

#117 Not as simple as it seems...

by sacolcor

Tuesday March 7th, 2000 8:53 AM

You are replying to this message

"I think any Webpage with a "valid" document type definition should be handled as a W3C compliant Webpage and all others should be handled as a non-W3C compliant Webpage."

That sentence makes it sound easy, but doing it is anything but. A "non-W3C compliant Webpage" isn't following any particular syntactical rules, but rather a hodge-podge of ill-defined (and sometimes contradictory) behaviors that have evolved through divergent browser development over the last five years. How is the browser to know whether to interpret a tag the way IE has done it or the way NS has done it? Even worse, what happens when a new web developer learns HTML/CSS/DOM by looking at a mix of web pages? They would end up mixing concepts from both camps and trying to use them in the same document.

Without a precise definition of the language, you'd have to build in an AI engine to try to make an intelligent guess on how to render the page, and even then I'd be skeptical of success.

I've decided to copy the post I made in the other thread, since it's germane to this topic: ---------------------------------------- I don't think many people understand the problems created by trying to support both the old way and the standard way. Let me mention a few of them: 1) Time. Would it be worth delaying the release of Mozilla by (I would guess) 2-3 months? Much of the old NS4.X code simply won't work in would all need to be rewritten.

2) Size. Adding the code to handle multiple rendering methods will bloat will be a bigger download, take up more memory, and run slower. Nokia's using this layout engine in their portable devices; this decision was undoubtably based in part on the fact that the code was relatively compact.

3) Ambiguity. There would be two (or more) possible ways to render many of the elements and calls. How does Mozilla know which one the designer intended? Some of the old ways of doing things conflict with what the standard ways, or with each other, and there will often be no way to tell.

4) Complexity. Adding this extra code, and especially trying to make decisions of which method to use, is going to make the code more complicated. This would incur a maintenance cost; trying to coordinate the two methods would require some pretty 'clever' code, which will undoubtedly create a number of 'clever' bugs that will need to be worked out.

All of this trouble, just to support techniques that there are generally new and better ways of doing just doesn't seem like a worthwhile thing to work on. I doubt that many web designers have been coding their sites for /only/ Naviagtor, in any case...they probably also coded for IE, and /most/ of the IE code I've seen looks pretty decent on Mozilla, since IE5 has something like 95% HTML4 and 85% CSS1. The ones I've talked to have generally been /eager/ to ditch their old NS4.X code ASAP, because it was filled with kludgy workarounds that were put in because of NS4.X's poor CSS support. ---------------------------------------- Including everyone in the W3C decision making process sounds good in theory, but breaks down in practice. Language design is one of those things that's not hard to do, but is very hard to do /well/. Making the decision making process open to the general public would result in slowing the process down to a snail's pace, as well-meaning people try to add a lot of features that sound nice, but really aren't a good idea. In this situation, a small, highly-skilled group of PhDs and industry experts can do the job better than a community consensus. And this process /is/ fairly open. Take XSL, for example. Discussion for it goes on the XSL mailing list, which you could subscribe to at <>. You can also comment on the working draft, by writing to <>. In all, the W3C is far more open than groups such as ANSI and ISO, where you can't even see a copy of their finished standards without forking over hundreds of dollars.