This repository is aim to create R parcel.
You need to install Docker. I confirmed with Docker community edition on Ubuntu.
After installation of Docker, run:
$ ./build.sh
You will find parcels in ./target
directory.
Prepare R environment using miniconda.
$ mkdir -p ./target/CONDAR-3.4.1/lib
$ cp -r ./source/meta ./target/CONDAR-3.4.1/
# (Option) Modify meta/parcel.json if needed
# $ sed -i s/__OS_VERSION__/el6/g ./target/CONDAR-3.4.1/meta/parcel.json
$ docker build -t chezou/condar:centos7 -f centos7.Dockerfile .
# Open another terminal and find your container.
$ docker run -it -t chezou/condar:centos7 /bin/bash
$ docker ps
$ docker cp -L <containerId>:/opt/conda/envs/R_env ./target/CONDAR-3.4.1/lib/conda-R
# Terminate docker process
In current version, you should modify version info in json.
$ cd target
$ tar cvzf CONDAR-3.4.1-el7.parcel CONDAR-3.4.1 --owner=root --group=root
I borrowed validator.jar and make_manifest.py from cm_ext repo.
See also: How to use validator.jar
$ java -jar lib/validator.jar <args>
At least, you should pass parcel test:
$ java -jar lib/validator.jar -f target/CONDAR-3.4.1-el7.parcel
Run following command, then you'll get ./target/manifest.json
.
$ python ./lib/make_manifest.py ./target
Upload every ./target/*.parcel
and ./target/manifest.json
to the same directory on http server.
You can put Parcels on S3. It doesn't need to have index.html for directory listing.
This parcel use Miniconda to install R. You can controll R version with setting r-essentials' version.