Platform

We deployed virtual machines in the cloud at VirtualData. To ensure that everyone observes the same system performance phenomena during the practicals, it is strongly suggested that you do them on the Linux server that you have been given access. All instructions given in this document will assume that you are doing so, and may not work as expected on another computer.

Network connexion

At least one of the academic institutions with which you work likely provides you with access to eduroam. Be sure to set it up and try it out in advance: you may need help from your local computing service as the configuration procedure is institution-specific. Here’s the documentation for IJCLab members.

In case you experience last-minute issues with eduroam, you will alternatively be able to use our public Wi-Fi network. You will need to be physically at IJCLab to set up this one, so no advance preparation is possible in this configuration.

Server configuration

The practical servers are configured in the following manner:

Each server is running an AlmaLinux OS (Alma 9).
The username to connect is student, and you should use the password given at the start of the lecture.
All necessary tools to compile and execute scripts are installed and accessible in the server. If this is not the case, please ask one of the teachers.
When you connect to the server, you are automatically put in a Python virtual environment (called robprog). This environment contains all necessary packages to perform the work. If you deactivated it, just log out from the server and reconnect, or execute source /opt/.venv/bin/activate.

Finally keep in mind that

You should not use the server for other purposes than this lecture.
The server will be available until the 12th April, then it will be reallocated to other work and all data will be removed. This should give you enough time to finish the second homework, but please remember to scp any file you care about away from the server!

SSH shortcuts

At the beginning of the course, you have been provided with the IP address of the server that you will be working on, as well as your user name on it. However, IP addresses are unwieldy to remember and user names are boring to specify on every connexion, so you may want to add something like the following to your SSH config file (~/.ssh/config on Linux and macOS)…

# Robust programming training
Host robust-programming
    User <username>
    HostName <server IP address>

…where you replace the <bits between brackets> with the matching information from the registration e-mail that you have received. Once that is done ssh robust-programming will connect you directly to the practicals platform, with the right user name, and directly ask you for your password.

Source code

First of all, you should download the source code for the practicals. It is located inside of this archive, which you can download and unpack in your home directory using this script:

#!/bin/bash

FOLDER_NAME="robust-programming"
BASE_URL="https://informatique-des-deux-infinis.pages.in2p3.fr/pheniics"

cd ~

# Keep a copy of the current
# folder if it exists
if test -d ${FOLDER_NAME}; then
  SUFFIX=`date +"%Y%m%d_%s"`
  BACKUP=${FOLDER_NAME}_${SUFFIX}
  echo "Found existing ${FOLDER_NAME} folder"
  echo "Keeping a copy under ${BACKUP} and redownloading it..."

  mv ${FOLDER_NAME} ${BACKUP}
fi
curl ${BASE_URL}/${FOLDER_NAME}/${FOLDER_NAME}.tar.gz | tar -xz

If, for some reason, you get one of these files into an undesirable state (e.g. accidentally delete it), you can get back a fresh copy of it in its initial state by re-running the script. It will save your current work, and download again the robust-programming directory in which you will find the original source files.

Eventually, your home directory ~/robust-programming directory containing the source code for all practicals. For demonstrations, this will be the final source code at the end of the demonstration. You can compile C++ programs using the provided Makefile:

make -j$(nproc)

Some Makefiles also have additional options, such as make test for running the test binaries (if any) or make doc for building Doxygen documentation. Please feel free to check the contents of the Makefiles for more info about what they do.

Local file access

One drawback of using an SSH-based remote configuration is that it seemingly forces you into using archaic terminal-based code editors like vi or emacs, instead of more modern alternatives like Sublime Text or Visual Studio Code. Thankfully, there are ways to mount a remote directory that you have access to via SSH as a local directory on your laptop, so that you can manipulate files in it locally with your preferred text editor.

For a general way to do this, please check out this sshfs tutorial that Hadrien wrote as part of the documentation of srv-calcul-ambulant, another compute server that he administrates. Translating these instructions for a different server should be straightforward (you only need to change the server name), but do get back to us if you encounter any issue.
If you are using Visual Studio Code specifically, it has its own support for operating over SSH connections that you may prefer over sshfs.

Of course, this access is only good for editing source code, and you should not try build programs or execute them on your local machine: you may not have the required prerequisites installed and even if you do, your local system is probably not as well tuned for reliable and reproducible performance benchmarking as the remote one.

Robust Programming