Re: A simple guy with a simple problem

[Jonathan Borden <jborden@mediaone.net>]

Sean McGrath wrote:

>
> Grove/Infoset approach: Possible to get the processing right
> and leave the document "the same". Serious complexity
> jump both in terms of the underlying abstractions to grasp and
> the coding required.
>
> My oft-repeated thesis is that I am loath to concede that
> the complexity of the grove/infoset approach is unavoidable.

All too often the terms "infoset" or "grove" are thrown about in a way that
suggests some incomprehensible complexity. In day to day practice, and
particularly for a "simple guy with a simple problem" I say this:

You have posed the wrong problem! Who cares if the document is identical?
SAX (or DOM or XPath or JDOM etc etc etc) provides an interface to handle
processing of your document. When using one of these common interfaces a
bunch of irrelevent information is thrown out. The "infoset" is just a fancy
term that says what information is relevent and what is irrelevent.
Basically this says that if you can't get the information from a SAX or DOM
interface then forget about it (remember you are a simple guy with a simple
problem).

> I am championing the complete separation of DTD and
> instance as a first step towards exploring an alternative,
> layered approach to this sort of processing which makes
> a parsing based paradigm workable whilst leaving
> the unaffected parts of a document "the same" en route to
> further processing stages.
>

Firstly simple guys with simple problems _don't_ use internal subsets. Just
because your toolset includes a fancy wrench doesn't mean that you ought use
it for tightening a simple nut. My advice is to put that wrench away for a
day in which your simple wrench doesn't work -- but don't throw it away
because the day you need it, you'll be glad its in your toolbox.

-Jonathan