We show here how to install Rumble from the github repository if you wish to do so (for example, to use the latest master). However, the easiest way to use Rumble is to simply download the already compiled .jar files.
The following software is required:
- Java SE 8+ (last tested on 8u221). The JDK is needed here, but the JRE is enough with the packaged jar file.
- Spark, version 2.4.3 (for example)
- Ant, version 1.10.1
- ANTLR, version 4.7 (supplied in our repository)
- Maven 3.6.0
Important: the ANTLR version varies with the Spark version, because Spark is also shipped with an ANTLR runtime (example: Spark 2.2.0 is with ANTLR 4.5.3, Spark 2.3.0 with ANTLR 4.7). The ANTLR runtime MUST match the ANTLR generator used to generate the Rumble jar file.
Checking the requirements
Type the following commands to check that the necessary commands are available. If not, you may need to either install the software, or make sure that it is on the PATH.
$ java -version $ mvn --version $ ant -version $ spark-submit --version
You first need to download the rumble code to your local machine.
In the shell, go to the desired location:
$ cd some_directory
Clone the github repository:
$ git clone firstname.lastname@example.org:Rumble/rumble.git
Go to the root of this repository:
$ cd rumble
For convenience, we have included the ANTLRv4 files for 4.5.3 and 4.7 in the lib directory of the repository (see corresponding license).
From the root directory of the rumble local checkout, you first need to build the parser:
$ ant -buildfile build_antlr_parser.xml generate-parser -Dantlr.jar=/lib/antlr-4.7-complete.jar
Once the ANTLR sources have been generated, you can compile the entire project like so:
$ mvn clean compile assembly:single
After successful completion, you can check the
target directory, which should contain the compiled classes as well as the JAR file
The most straightforward to test if the above steps were successful is to run the Rumble shell locally, like so:
$ spark-submit --master local --deploy-mode client target/spark-rumble-1.1-jar-with-dependencies.jar --shell yes
The Rumble shell should start:
Using Spark's default log4j profile: org/apache/spark/log4j-defaults.properties ____ __ __ / __ \__ ______ ___ / /_ / /__ / /_/ / / / / __ `__ \/ __ \/ / _ \ / _, _/ /_/ / / / / / / /_/ / / __/ /_/ |_|\__,_/_/ /_/ /_/_.___/_/\___/ Master: local Item Display Limit: 1000 Output Path: - Log Path: - Query Path : - jiqs$
You can now start typing interactive queries. Queries can span over multiple lines. You need to press return 3 times to confirm.
jiqs$ "Hello, world!"
This produces the following results (
>>> show the extra, empty lines that appear on the first two presses of the return key).
jiqs$ "Hello, world!" >>> >>> Hello, world
You can try a few more queries.
jiqs$ 2 + 2 >>> >>> 4 jiqs$ 1 to 10 >>> >>> ( 1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
This is it. Rumble is step and ready to go locally. You can now move on to a JSONiq tutorial. A Rumble tutorial will also follow soon.
Running on a cluster
You can also try to run the Rumble shell on a cluster if you have one available and configured -- this is done in the same way as any other
$ spark-submit --master yarn --deploy-mode client --num-executors 40 spark-rumble-1.1-jar-with-dependencies.jar
More details are provided in the rest of the documentation.