Re: [Gnu-arch-users] Re: documentation as info

[Tupshin Harper]

Florian Weimer wrote:

> Tupshin Harper <address@hidden> writes:
>
>  
>
> > Or use an editor that removes the need to interact with the
> > verbosity. The lack of contraction features is one reason that XML has
> > proliferated in a way that SGML never did. It makes it *much* easier
> > to write an XML parser, and also makes the parsing much faster. A
> > worthwhile tradeoff.
> >    
>
> No, it's not.  Computers are quite good at parsing, and even SGML is
> not that hard (even though it's a bit bloated).
Performance is relative. The lack of optional tags *does* make parsing 
faster. Sometimes, this is important.

>  The lack of optional
> tags makes it either hard to write uniform parsers (because the XML
> DTD does not adequately describe the document structure)
This doesn't make sense at all. Clearly you mean implied tags, and not 
optional tags which XML has. The lack of implied tags doesn't impair the 
ability to describe document structure at all.

> or to write
> the documents using typical *NIX tools (because all the structure has
> to be made explicit using tags).
>  
Again, the lack of implied tags requires additional verbosity, but it 
isn't harder to generate using any procedure I can think of.

> I've written down a few thoughts on this dichotomy, when I designed
> the XML DTD for my web pages:
>
>  <http://www.enyo.de/fw/notes/xml-dtd-design.html>
>  
Interesting read...I agree with some bits, and strongly disagree with 
others. I applaud your using XML as a source format and transforming to 
HTML. Some thoughts:
* You complain about XML's need for a BODY wrapper, but correct HTML 
requires it as well.
* You show how DTDs have inherent redundancy, but never mention that 
those redundancies don't need to exist in XML schemas which are a vast 
improvement over DTDs.
* You never mention the negative aspect of implied tags which is that it 
is *much* easier to make an undetected mistake.
* You correctly identify dates as something that is cumbersone to 
represent structurally in XML. In general there is a collision between 
the verbose markup and the desire to structurally represent very small 
pieces of information. It would be great if you could define an XML 
attribute as a delimited (dashes, in this case) list of fields, each 
having it's own symantic meaning. I don't know if anybody has proposed 
this to W3C, but if not, they certainly should.

-Tupshin