also... SGML/DTD

[William Shipley]

>  What exactly would be the benefit of restricting yourself
>  from viewing the nearly countless number of pages on the
>  web which do not conform precisely to the latest DTD?

Actually, I think the idea is that we would be able to parse any Standard  
Generalized Markup Language file if given a DTD for it.  HTML is an SGML,  
but since the original browsers had incredibly simple parsers (that didn't  
understand true SGML), people have been abusing HTML to the point that it's  
no longer acceptable to just have a parser that understands the DTD; we also  
have to understand and try to correct all the errors people generate.

Imagine if the C language writers had said, "Here's the language, but,  
really, you have to account for and correct a whole bunch of errors that the  
users may make," rather than just returning an error when the user makes an  
error.  Well, compilers would be a whole lot harder to implement, which  
would restrict the number available and make them all suck a little harder,  
since compiler writers would have to spend their time write error-correcting  
code instead of concrentrating on making a good compiler.

All that said, OmniWeb 2.0 will probably be a lot more intelligent about  
error correction, since that's what the market needs.  Understanding true  
SGML with an arbitrary DTD is basically a separate problem from writing a  
web browser, and although I think it's an interesting one I don't see as  
much market for it.  I'd rather be adding other cool features.

-Wil