IQ packet without 'id' attribute causes connection drop

CenterIM Bug-Report
http://bugzilla.centerim.org/show_bug.cgi?id=119

Miranda Bug-Report
http://bugs.miranda-im.org/view.php?id=501

IMHO not kicking either side is the best solution usabiliti-wise if not standards-wise.
This patch seems to work for us:

— src/java/org/xmpp/packet/IQ.java 2008-11-05 16:16:16 +0000
+++ src/java/org/xmpp/packet/IQ.java 2008-11-16 00:35:37 +0000
@@ -67,6 +67,10 @@
*/
public IQ(Element element) {
super(element);
+ if (getID() == null) {
+ String id = String.valueOf(random.nextInt(1000) + "-" + sequence++);
+ setID(id);
+ }
}

/**
@@ -80,6 +84,10 @@
*/
public IQ(Element element, boolean skipValidation) {
super(element, skipValidation);
+ if (getID() == null) {
+ String id = String.valueOf(random.nextInt(1000) + "-" + sequence++);
+ setID(id);
+ }
}

/**
@@ -95,6 +103,10 @@
// Copy cached JIDs (for performance reasons)
this.toJID = iq.toJID;
this.fromJID = iq.fromJID;
+ if (getID() == null) {
+ String id = String.valueOf(random.nextInt(1000) + "-" + sequence++);
+ setID(id);
+ }
}

/**

Ok, gave it a quick test, looks good. However, I suggest a default value of true instead of false for the new system property xmpp.server.validation.enabled

If packets will still be able to get into the server then we need to fix the known NPE besides fixing StanzaHandler. However, we should know that other NPE could still happen since IQ packets may go to components that will have the same problem.

What about doing that same that the plugin does? (simply adding that ID attribute)

However, best way would be solving this issue on client side.