- From: Joseph Kesselman <keshlam@us.ibm.com>
- Date: Tue, 27 Mar 2001 16:45:27 -0500
- To: WWW DOM <www-dom@w3.org>
>>Every implementation of W3C's Document Object Model will include their own >>extensions to the implementation. Maybe I'm missing the point, but it seems to me that there's less to this statement than meets the eye. There's a very long tradition of "mining" code for useful undocumented methods. Of course doing so means sacrificing portability -- often including portabilty to future releases of that same package. There's also a long tradition of implementation-specific documented extensions. Some are filling gaps in the APIs, or proposing/prototyping functionality which may be added to the API in the future. Some are adding behavior specific to that particular application set. Again, portability is a concern. >programmers find that in order to get the job done they need more that >what the Recommendation calls for. Sometimes, yes; the DOM Level 2 Recommendation doesn't yet provide a way to discover whitespace-in-element-context, for example. Sometimes it's a convenience/performance feature, and someone happened to decide to implement it directly on the object rather than on a helper object. And again, the portabilty issues are self-evident. None of this is DOM-specific, and I don't find any of it surprising. Certainly we should keep an eye on what's being done in the way of extensions, and consider whether any of them are general/important enough that the DOM ought to pick them up. But if folks insist on coding against specific implementations, we can't stop them... and I'm not sure we ought to try; this might actually be the right trade-off for their intended problem domain. We should probably advise them to encapsulate calls to this custom code, so they can reimplement this isolated function should they move to a different DOM... but I don't see a way to do more than that, other than to simply continue DOM development. ______________________________________ Joe Kesselman / IBM Research
Received on Tuesday, 27 March 2001 16:45:32 UTC