[
Lists Home |
Date Index |
Thread Index
]
[Joe English]
>The AFDR, if viewed as a transformation language, is extremely
>lightweight; there's not much you can do with it. Although
>it could be used to turn a full newsfeed into an RSS channel
>or DocBook into HTML (but not the kind of HTML you'd want
>to build a web site out of), it's much more useful as a
>design tool.
>
>The basic idea is, instead of writing software to work on
>a particular concrete document type, write it as an architectural
>processor; most of the time this just means keying off of #FIXED
>attributes instead of element type names. Then it can work
>with any concrete document type that conforms to the architecture;
>most of the time this can be accomplished just by adding a
>few #FIXED attributes to the DTD.
So, we could proceed as follows:
1) Parse in the normal fashion using schema language du jour
2) Map using a lightweight mapping utility, the concrete element types to
those of the required base classes
3) Key off these new element type names for processing
Works for me. Simple and Elegant with no frightening works like "meta" and
"abstractions". IT
appeals to the engineer in me:-)
All we need now is a pipeline processor to make joining these three phases
together a
trivial configuration exercise. A sort of assembly line maybe.
Hmmm...
Sean
http://xpipe.sourceforge.net
|
