Skip to content

Latest commit

 

History

History
189 lines (131 loc) · 6.57 KB

INSTALL.md

File metadata and controls

189 lines (131 loc) · 6.57 KB

Installation Guide

I-Test currently supports two programming languages, Python and Java.

We provide two options to install the required software for running I-Test and our scripts:

  1. use docker
  2. install to local environment.

Use Docker

We provide a Docker image pre-installed with all required software (conda, sdkman, texlive) and is ready to be used to replicate all experiments in this artifact. Please ensure at least 20GB of disk space to download the image and run experiments.

If you haven't, please install Docker Engine following the instructions here: Install Docker Engine.

Then, you can download our image by running

docker pull pipiyuyu/inlinetests:latest

You can start the container by running

docker run -it -m 4g pipiyuyu/inlinetests /bin/bash

This will drop you into the docker image's terminal at /home/itdocker. All content of this repo (on the latest release version) is located there.

To run something using sudo, the password is itdocker

Install to Local Environment

We utilize the package management systems (conda for Python and sdkman for Java) to install the necessary dependencies for I-Test itself, for our experiment scripts, and for running the unit tests of other open-source projects in our integrated experiments.

This document will guide you through the following steps (with some usage examples to verify if the installation is successful):

  • install the package management systems
  • install the I-Test framework
  • install the environment for running our experiment scripts

System Requirements

The minimum requirements for replicating our experiments are:

  • a Linux operating system (MacOS not guaranteed to work)
  • at least 10GB of free disk space

Note that I-Test framework itself is not limited to Linux; we have tested the I-Test framework on a MacOS machine (with MacOS 10.15.7). However, a part of our experiment scripts uses Bash and relies on Linux-specific grammars.

For your reference, we used the machine with the following specs to run experiments:

  • Intel Core i7-11700K @ 3.60GHz (8 cores, 16 threads) CPU
  • 64 GB RAM
  • Ubuntu 20.04 operating system

Installing Package Management Systems

The two package management systems we use, conda and sdkman, can both be installed in user mode (i.e., does not require sudo). To install each of them, you will need to execute some commands AND configure your .bashrc appropriately, and then restart the terminal for the changes to take effect. If you happen to have existing installations of (a recent version of) either package management system, you do not need to install it again.

Note: you don't have to install Python and Java on your own, as our scripts will automatically install the correct version of them using conda and sdkman.

conda for Python

  1. Download the Miniconda installation script from here; pick the correct link according to your CPU architecture (usually "Miniconda3 Linux 64-bit", but for ARM should be "Miniconda3 Linux-aarch64 64-bit"). For example:
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh

(Installing Anaconda is also fine, which is bundled with some libraries that we don't need.)

  1. Execute the downloaded script, and follow the prompts on the terminal to install conda at the desired location. For example:
bash Miniconda3-latest-Linux-x86_64.sh

  1. If not already done at the end of the last step, run conda init which automatically changes your .bashrc file to use conda.

  2. Restart your terminal. By executing conda --version, you should be able to see the version of conda you just installed.

sdkman for Java

  1. Execute the following command, and follow the prompts on the terminal to install sdkman.
curl -s "https://get.sdkman.io" | bash
  1. Restart your terminal. By executing sdk version, you should be able to see the version of sdkman you just installed.

Installing the I-Test Framework

Python 3.6+

  1. Change directory to "python". Assuming you were at the root of this repository: cd python

  2. Execute ./prepare-conda-env.sh

  3. Execute conda activate inline-dev

  4. You should be able to see "(inline-dev) " as the prefix of the prompt in your terminal. Then, to further check if installation is successful, you can run the tests for the I-Test framework in Python: pytest

  • Trouble shooting: if you get error "CondaEnvironmentError: cannot remove current environment. deactivate and run conda remove again", please run conda deactivate to exit the inline-dev environment, then try again.

Java 8

  1. Change directory to "java". Assuming you were at the root of this repository: cd java

  2. Execute ./install.sh

  3. During the installation in the previous step, we actually already run the tests for the I-Test framework in Java. You can also run these tests again to double-check if the installation is successful: sdk use java 8.0.302-open; sdk use maven 3.8.3; mvn test

Installing the Environment for Experiment Scripts

  1. Change directory to "research". Assuming you were at the root of this repository: cd research

  2. Execute ./prepare-conda-env.sh

  3. Execute conda activate inline-research

  4. You should be able to see "(inline-dev) " as the prefix of the prompt in your terminal. Then, to further check if installation is successful, you can run the script for running the inline tests in the 50 Python example files and 50 Java example files:

python -m research.exp_standalone run --language=python --requests_file=../data/exp/standalone/python.yaml --run_dir=../data/examples/python --out_dir=/tmp/inlinetest-smoke/python --force=True
python -m research.exp_standalone run --language=java --requests_file=../data/exp/standalone/java.yaml --run_dir=../data/examples/java --out_dir=/tmp/inlinetest-smoke/java --force=True

For both of the previous commands, you should see a progress bar that ends at "running: 100% ... 50/50 ...", without warning or error messages.

  • Trouble shooting: if you get error "CondaEnvironmentError: cannot remove current environment. deactivate and run conda remove again", please run conda deactivate to exit the inline-dev environment, then try again.