On Monday, 08/26/2002 at 10:50 MST, nirmalts <[EMAIL PROTECTED]> wrote: > If i give XML file as input, does Xalan make a DOM > tree out of it and query it with XPath expression.
Do you mean Xalan-J or Xalan-C? Xalan-J uses DTM rather than DOM as its internal model. See http://xml.apache.org/xalan-j/dtm.html for background informatin. > the whole tree kept in memory or does it use any > iterartive method to query the tree? For Xalan-J: See http://xml.apache.org/xalan-j/dtm.html#settings, specifically the Incremental option. Also see archives of this mailing list for discussion of "pruning", our term for discarding portions of the source tree which are no longer needed. This turns out to be a significantly difficult problem, given XSLT's generality and some of the trade-offs involved in DTM's design. We now have a "tail-pruning" solution working in a specific limited situation (using one DTM to contain multiple result tree fragments); "in our copious spare time" we hope to be able to start applying this to input documents as well. Early versions will probably operate as user-invoked extensions, though we do hope to automate this at a later date. > I have huge XML documents. I want to know if i should > write my own memory management routines or will Xalan > take care of it? I'm not sure "memory management routines" will suffice, for the same reasons that pruning is difficult. If you can break the document up into smaller pieces and operate on them independently -- not always possible, depending on what your stylesheet wants to do -- that could be worth doing. (Alternatively, start with smaller documents and merge them after they've been styled...) ______________________________________ Joe Kesselman / IBM Research
