Thoughts on XML authoring

Since moved nodesets over to my own website, I'm now authoring atom 0.8 in emacs, with namespaced elements. This one is currently <x:p> paragraph. This is the first time I've authored in this way, and it's not easy, or at least not as easy as it is without namespaces. When I first read the namespace rec, I took away this simplistic impression that namespaces were inherited. I thought that if I opened a document in namespace a, specifying the namespace on that element, then that namespace would remain in force until I changed it. Too simplistic? Perhaps, but think how easy it would have been to write using such an idea

<a xmlns='namespace 1'>
<b>this would be in ns1
   <c>This too</c>
</b>
<d xmlns='namespace 2>
  But this would change the ns
  <e>And this would be in namespace 2</e>
</d>
</a>

I'm not sure if this is clear, but the intent should be. The namespace associated with an element is the one declared in the nearest ancestor. The lack of a namespace declaration is immaterial. If namespaces are in use in a document, then every element is within at least one namespace. Namespaces declared in ancestors of the nearest one are less closely associated, hence processors can ignore them.

The advantages I see in this form of editing are that the author is freed from using prefixes, which I guess is what I'm railing against, the editorial overhead of writing them. The disadvantage is that I need to repeat the actual namespace each time I use it, which leaves more room for error than typing a prefix. I guess some bright spark could come up with a syntax that works around that. My goal is to remove the need to write them.

End of rant

test file for atom content generation

For some time now, I've been using plain text as a way of writing notes to myself, to remind me of something, to keep a note of a useful url etc. Believing I'd have a use for these in XML, I wrote some Python to convert these to XML. Thinking of my last entry http://www.dpawson.co.uk/nodesets/entries/050627.htmland reading (almost inadvertently) a piece by Martin Fowler on mini languages, http://martinfowler.com/articles/languageWorkbench.html#ASimpleExampleOfLanguageOrientedProgrammingI started to think about content generation in plain text. Looking at the xhtml body content of an entry, I started by just processing that part. So I'm writing this one in plain text, and processing it to get output in xhtml, with a wrapper of <x:div> such that it can be placed directly within the <a:content> element At a basic level, it all seems to be working.

Changing the subject.

Or not, in this case. Am I the only one that feels that XML markup is a bit of a waste of time at the detail level? Whilst still being absolutely convinced that XML is right as a data interchange format, as a format for retaining documents over a long period, and that complex structures really should be manually marked up by hand; I'm not in the least convinced that I'm any better than my machine at generating pointy brackets in the right place. I won't speculate just where the break point is, but for docheads, I'm sure machine processing of plain text is more of a help than a hinderance.

Now lets see if I can slot this content into another blog entry.