This configuration we are talking about, performance differences won't matter much IMO.
G <div>-------- Original message --------</div><div>From: Matt Sicker <[email protected]> </div><div>Date:05/11/2014 20:02 (GMT-05:00) </div><div>To: Log4J Developers List <[email protected]> </div><div>Subject: Re: Delivery Status Notification (Failure) </div><div> </div>SAX or StAX could be faster, especially since we're not using the DOM per se but a parallel in-memory DOM-like tree. On 11 May 2014 18:46, Ralph Goers <[email protected]> wrote: I can't imagine what would be faster than traversing the DOM as we currently are. Sent from my iPhone On May 11, 2014, at 11:46 AM, Matt Sicker <[email protected]> wrote: JAXB wouldn't be as flexible as the current XML format, that's for sure. I've also read on FasterXML's site that Woodstox is supposed to be faster than Jackson with XML, but I haven't read much into that. On 10 May 2014 22:24, Ralph Goers <[email protected]> wrote: The reason I did the XML configuration the way I did was because it was easy to map the Document to Nodes. I thought about using JAXB but it didn't seem like a good fit. I didn't know about Jackson at the time but I am not sure it would have been easier. Sent from my iPhone On May 10, 2014, at 7:45 PM, Gary Gregory <[email protected]> wrote: FWIW, I am redoing the JSON and XML layouts to use the same Jackson code as for reading them back in for the listener side. In my impl, the context map is a IO'd as a map and the context stack as an array. As you may have seen despite the mail outage, I've been committing in chunks for the pieces I'll need before the new JSON/XML Jackson-based IO. Jackson is nice for that, we get 2-for-1, which make me wonder why we do not do our config reading using Jackson for XML. I suppose XML was impl'd before JSON and YAML. Gary On Sat, May 10, 2014 at 4:25 PM, <[email protected]> wrote: > Author: rgoers > Date: Sat May 10 20:25:28 2014 > New Revision: 1593727 > > URL: http://svn.apache.org/r1593727 > Log: > LOG4J2-623 - Format ThreadContext properties as a JSON map in JSONLayout > > Modified: > > logging/log4j/log4j2/trunk/log4j-core/src/main/java/org/apache/logging/log4j/core/layout/JSONLayout.java > > logging/log4j/log4j2/trunk/log4j-core/src/test/java/org/apache/logging/log4j/core/layout/JSONLayoutTest.java > logging/log4j/log4j2/trunk/src/changes/changes.xml > > Modified: > logging/log4j/log4j2/trunk/log4j-core/src/main/java/org/apache/logging/log4j/core/layout/JSONLayout.java > URL: > http://svn.apache.org/viewvc/logging/log4j/log4j2/trunk/log4j-core/src/main/java/org/apache/logging/log4j/core/layout/JSONLayout.java?rev=1593727&r1=1593726&r2=1593727&view=diff > > ============================================================================== > --- > logging/log4j/log4j2/trunk/log4j-core/src/main/java/org/apache/logging/log4j/core/layout/JSONLayout.java > (original) > +++ > logging/log4j/log4j2/trunk/log4j-core/src/main/java/org/apache/logging/log4j/core/layout/JSONLayout.java > Sat May 10 20:25:28 2014 > @@ -253,34 +253,25 @@ public class JSONLayout extends Abstract > buf.append(','); > buf.append(this.eol); > buf.append(this.indent2); > - buf.append("\"Properties\":["); > - buf.append(this.eol); > + buf.append("\"Properties\": {"); > final Set<Entry<String, String>> entrySet = > event.getContextMap().entrySet(); > int i = 1; > for (final Map.Entry<String, String> entry : entrySet) { > - buf.append(this.indent3); > - buf.append('{'); > buf.append(this.eol); ----- Message truncated ----- -- E-Mail: [email protected] | [email protected] Java Persistence with Hibernate, Second Edition JUnit in Action, Second Edition Spring Batch in Action Blog: http://garygregory.wordpress.com Home: http://garygregory.com/ Tweet! http://twitter.com/GaryGregory -- Matt Sicker <[email protected]> -- Matt Sicker <[email protected]>
