How to start working with Sedona

What do you need to install ?

Easiest way, create a container to work inside

You can use the Dockerfile in the repo If you haven't docker, install it:

sudo apt-get install docker

Then you will be able to create a iso image. You could have more details https://phoenixnap.com/kb/create-docker-images-with-dockerfile

docker build -t <name you choose for the image> path/to/Dockerfile

After that, you can check if you created an iso image successfully by running the following command:

docker images

You must found something like this:

REPOSITORY    TAG        IMAGE ID       CREATED         SIZE
boussole      v3         91e4191c2521   4 weeks ago     2.34GB
boussole      v1         e4d216782d8a   2 months ago    1.79GB

In my case, I do Geomatics so I originally choose boussole who means compass in french. Do as you prefer.

Now, you have to start your docker. You can use the following:

docker run -it -v $PWD:/path/to/your/host/machine/working/directory/boussole/ boussole:v1 bash

You could find explanations https://www.howtogeek.com/733522/docker-for-beginners-everything-you-need-to-know/ but the idea is to start your container in you working directory in order to safe all you created inside your container in your host machine. DOn't hesitate touse https://excalidraw.com/ to understand what you're doing. Therefore, you must see something like this:

root@8a61c067bf09:~/projects/boussole#

Clone Sedona

Now, we have working environment. Let's use it to start working with Sedona. But before, we need the Sedona code. To do so, start by forking the official repo. Go https://github.com/apache/sedona. At the upper right side of the page, you will found a Fork button. Click on it and select "New fork" or something like that. Congratulations, you have your own fork of the Sedona project. Now, clone it in your host machine

[email protected]:apache/sedona.git

Create your so-called fat jar

Go in Examplesdirectory of the Sedona source code. There, run:

mvn clean package

This command will take the examples codes and create a .jarbinary file after compiling it.

Submit your fat jar

Go in /target/ directory and run your spark-submit file which could seems like this:

#!/bin/bash

$SPARK_HOME/bin/spark-submit \
  --master local[2]\
  --driver-memory=2g\
  --executor-memory=2g\
  --packages org.apache.sedona:sedona-spark-shaded-3.0_2.12:1.6.0,org.datasyslab:geotools-wrapper:1.6.0-28.2\
  --class Main original-sedona-spark-example-1.0.0.jar

Nota bene: according if you want to launch an example which use your modified Sedona code, you will need to change this file and the pom.xml file to explain to the compiler what piece of code does it has to take into account. Therefore, you will just launch this .sh file:

./spark-submit.sh

Now, you must have launched your first Sedona example. You can now retry with a modified version !