Parsing from InputStream problem

Question

I have a serious problem with parsing XML from an InputStream.
It is the exact same problem other people had before in those threads :
http://forum.java.sun.com/thread.jsp?forum=34&thread=343710
http://forum.java.sun.com/thread.jsp?forum=34&thread=68180
http://forum.java.sun.com/thread.jsp?forum=34&thread=68259
I have read the FAQ at http://www.jdom.org/docs/faq.html which suggest the following :
byte[] buf = new byte[length];
new DataInputStream(inputStream).readFully(buf);
InputStream in = new ByteArrayInputStream(buf);
(Contributed by Joseph Bowbeer)
But my problem still remains the same. What happens if more than 1 XML documents were writen in the InputStream because the writting took place very fast and the receiver could not parse the XML document fast enough so now the byte buffer 'buf' contains 2 XML documents one after the other ? What if it contains half XML document because the sender was temporarily blocked or something; and after some time it continues with the rest XML document ?
how could I overcome those problems ?
thanks

Answer 1

The root cause of the problems reported in this and other threads seems to be the SAX parsers are buffering the entire input stream before they commence parsing (which appears to be completely contrary to the spirit of SAX). This explains the parser hanging behaviour reported by some folk. It also explains why the parsers are awaiting for some sort of EOF marker before they begin parsing.
Maybe this buffering behaviour has been fixed in the newer releases of xerces and crimson et al, but there is a good discussion of this problem and a solution (albeit for xerces), at http://xml.apache.org.xerces-j/faq-write.html, under the section "how do I read data from a stream as it arrives?".