- From: Oleg Tkachenko <oleg@tkachenko.com>
- Date: Sat, 31 Jan 2004 17:23:35 +0200
- To: Elliotte Rusty Harold <elharo@metalab.unc.edu>
- Cc: daniel@veillard.com, www-xml-xinclude-comments@w3.org
Elliotte Rusty Harold wrote: > The big problem with eliminating streaming as a viable implementation > possibility is that you limit the maximum size of documents that can be > handled. XInclude is probably good up to what I consider a medium sized > document such as a book. However, for truly large documents, XInclude > becomes infeasible if it's necessary to store all documents in memory > until processing is complete. > > It's also worth noting that several existing streaming implementations > such as the XInclude handler in Xerces are basically subsetting the spec > by eliminating support for document fragments. The question is whether > such subsetting should be blessed or discouraged. Well, in XInclude.NET we are kinda subsetting too, not supporting intra-document refs. In fact XInclude.NET is always streaming for main document and for included one it's streaming or DOMing depending on presence of XPointer. That might change in the future releases and main idea is to keep streaming, but in the face of intra-document refs and XPointer it's really questionable. -- Oleg Tkachenko http://blog.tkachenko.com Multiconn Technologies, Israel
Received on Saturday, 31 January 2004 10:24:45 UTC