Subject: RE: [xsl] XPath for matching multiple child elements From: "Michael Kay" <mike@xxxxxxxxxxxx> Date: Fri, 29 Sep 2006 17:39:31 +0100 |
> On the other hand, I'm not sure pipelining in itself is any more (or > less) "natural" than template matching; Yes, but the point about maintainability isn't that the design is necessarily easier to establish in the first place, it's that it's open to change and resistant to structural decay once in place. In particular, new functionality can often be introduced by adding a pipeline component, which means there is minimal need to understand the internal logic of existing components, and minimal risk of destabilizing them. The pipeline I use internally in the Saxon serializer has been a great success from that point of view. Spec changes like doing URL escaping before rather than after character substitution would have been a nightmare with a monolithic approach; as it is, they just involve a small change to the component that assembles the pipeline. And although there's an overhead in passing every node through a pipeline of a dozen processing steps, many of which might pass it through unchanged, this is offset by the fact that the pipeline is configurable to only apply those processes that are actually needed, so a conditional like "if XML 1.1" or "if normalizing" is only being applied once at pipeline-configuration time, rather than once for each node (or in the worst case, each character!) processed. Michael Kay
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
RE: [xsl] XPath for matching multip, Wendell Piez | Thread | Re: [xsl] XPath for matching multip, Will McCutchen |
RE: [xsl] XPath for matching multip, Wendell Piez | Date | Re: [xsl] move tag up to level, Mukul Gandhi |
Month |