Subject: Orthoganlity Questions From: Paul Prescod <papresco@xxxxxxxxxxxxxxxxxxxxxxxxx> Date: Fri Mar 28 23:52:20 1997 EST |
Often the design decisions that we do not understand about a language indicate the parts of that language that we fundamentally do not understand. Other times they are "shortcuts" (for performance, backwards compatibility, typing convenience, whatever). Occaisionally they are just oversights. Here are a few things that are bothering me a little about DSSSL: Why can't element construction rules be at least as powerful as match-element? and process-matching-children? Guess: implementation simplicity. Why is there no "or" or "not" for match-element? There is a very coarse-grained "or" where you can duplicate an entire pattern, but not within patterns. Guess: implementation simplicity. Why is there no (process-matching-descendants)? Sure, it is easy to write. But still -- so is (process-children). I have to introduce the concept of a node-list into my tutorial "early" just to implement this one function. node-lists are important and useful, but there is an order to learning things. This is probably not an oversight. So I probably just don't understand the good reason it doesn't exist. Does anyone know or want to guess? Paul Prescod
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
Re: DTD for customizable stylesheet, James Clark | Thread | Re: Orthoganlity Questions, James Clark |
Re: Design for (interactively) cust, Paul Prescod | Date | Re: Design for (interactively) cust, James Clark |
Month |