The project checked out via SVN builds 5 jar files, including "json.jar" and "commons-logging-1.1.1.jar". I am a little bit confused, as : 1. The default AppEngine? SDK includes a "repackaged-appengine-commons-logging-1.1.1.jar" 2. The default jars also include a repackaged JSON library

Which one to import in the user code ?

How do we kick off a mapper from cron? The example of programmatically creating a Configuration seems totally useless - it just shows how to make a web form that starts the job. We need some way to kick off the job from code!

For those wondering about the dependencies (jars); they are either downloaded or are included in the source download from svn and built with ant.

http://demofileuploadgae.appspot.com/ - My Appspot Mapper Reduce, Blobstore Demo

Java Import Mapper Code: Source

@lhoriman You can use the TaskQueue? API to start jobs programatically, i.e. in response to someother user action or via a cron job

	/**
	 * Starts a map/reduce job that updates {@link Subscriber} entities with 
	 * dates from a given status snapshot.
	 */
	public static void updateSubscriberDatesFromSnapshot(UsageBlob statusBlob) {		
		
		TaskOptions task = buildStartJob("Update Subscriber Dates");
		addJobParam(task, StatusBlobInputFormat.SAMPLE_ID_PARAMETER, statusBlob.getSampleId());
		addJobParam(task, StatusBlobInputFormat.DATE_PARAMETER, statusBlob.getDate().getMillis());
		
		Queue queue = QueueFactory.getDefaultQueue();
		queue.add(task);
	}


	private static TaskOptions buildStartJob(String jobName) {
		return TaskOptions.Builder
		.url("/mapreduce/command/start_job")
		.method(Method.POST)
		.header("X-Requested-With", "XMLHttpRequest") // hack: we need to fix appengine-mapper so we can properly call start_job without need to pretend to be an ajaxmethod
		.param("name", jobName);
	}
	
	private static void addJobParam(TaskOptions task, String paramName, String paramValue ) {
		task.param("mapper_params." + paramName, paramValue);
	}
	
	private static void addJobParam(TaskOptions task, String paramName, long value) {
		addJobParam(task, paramName, Long.toString(value));
	}

this is great... nice work! one thing ... have you consider to move to maven as a build tool? I may even give you some help if you want thanks Luigi

Hello, Is "Defining the Descending Key Index" still necessary? App Engine gives error log as shown below. This index is not necessary, since single-property indices are built in. Please remove it from your index file and upgrade to the latest version of the SDK, if you haven't already.

No, it is not necessary.

Yeah, it's no longer necessary. Just removed it from the documentation.

Another approach to auto-start MapReduce? with Blobs. No need to create mapreduce.xml file anymore :)

public static void StartMapReduceWithBlob(Class<? extends AppEngineMapper<BlobstoreRecordKey, byte[], NullWritable, NullWritable>> blobMapperImpl, String blobKey) {
	Configuration conf = new Configuration(false);
	conf.setClass("mapreduce.map.class", blobMapperImpl, Mapper.class);
	conf.setClass("mapreduce.inputformat.class", BlobstoreInputFormat.class, InputFormat.class);
	conf.set(BlobstoreInputFormat.BLOB_KEYS, blobKey);
    
	String xml = ConfigurationXmlUtil.convertConfigurationToXml(conf);
        
	Queue queue = QueueFactory.getDefaultQueue();
	TaskOptions task = TaskOptions.Builder
			.withDefaults()
			.url("/mapreduce/start")
			.method(Method.POST)
			.param("configuration", xml);
        queue.add(task);
}

Actually mapreduce.xml file is not needed in sandbox, but AppEngine? raises an error if it does not exist. The file can contain only a root element <configurations></configurations> in order to work.

private static TaskOptions? buildStartJob(String jobName) {

return TaskOptions?.Builder .url("/mapreduce/command/start_job") .method(Method.POST) .header("X-Requested-With", "XMLHttpRequest") // hack: we need to fix appengine-mapper so we can properly call start_job without need to pretend to be an ajaxmethod .param("name", jobName);

How do we get parameter "name" in map() function of map reduce class? Thanks

I have built on my PC. Please download the dist folder from http://goo.gl/RQ9ln