WS22/23 ULG Data Science

Package manager

We have already seen the Julia package manager. In the following, we will discuss this feature in more detail. We already know how to open the package manager: Simply press ]. Now, in order to, for example, add the Revise package type

(@v1.8) pkg> add Revise

From now on, every time we open Julia, we can type

julia> using Revise

which in this case adds functionalities to reload functions/modules if their definitions change. In case we wish to remove this package again we can open the package manager and type

(@v1.8) pkg> rm Revise

If we now want to install a specific version of a package we can add @<versionnumber> behind the package name. As an example, to install Revise at version $3.4.0$ we type

(@v1.8) pkg> add Revise@3.4.0

Or in order to install the latest development version

(@v1.8) pkg> add Revise#master

which used git to get the latest commit from the master branch. Moreover, we can update all packages that we have added by typing

(@v1.8) pkg> update

or just a specific package (for example Revise) by typing

(@v1.8) pkg> update Revise

To get a list of all installed packages we can type

(@v1.8) pkg> status

It is good practice to add packages only for certain projects to prevent conflicting packages, see dependency hell. This can be done by creating a new package environment. Here we create an environment called example:

(@v1.8) pkg> activate --shared example

If we check this new environment with the status command we will see that we have a clean package environment. We can return to the default environment by typing activate. We type activate ., if we want to create an environment that is identified with the project folder we are currently working in. Whenever someone uses this project folder, this person can enter the same environment by typing activate . and install all needed packages with the used versions by typing instantiate. The installed packages and the exact versions are saved into a Project.toml and a Manifest.toml file. The Project.toml file is also used by packages in order to define its version, dependencies and compat bound for dependencies.

If we do not want to start the REPL we can still include the environment by simply typing

julia --project=. main.jl

Note that packages installed in the standard environment ((@v1.8)) can be used in any other environment. Thus it is good practice to install packages for development (e.g. Revise) only in the standard environment and not in the project specific environment.

Let us practice the learned concepts a little bit.

Create a new project called ULG. To setup the project, create a new folder and navigate to it using the REPL.
Make sure you can use the Plots package in the new project environment by loading it with using Plots.
Now you can for example plot the connection between two points $(1,1)$ and $(2,3)$ by plot([1; 2], [2; 3]).
Check if the Plots package is available in the standard environment.
Install the latest version of the Revise package in the standard environment and version 3.3.0 in your project environment.

Solution

shell> mkdir ULG

shell> cd ULG
/home/stephan/projects/ULG

(@v1.8) pkg> activate .
  Activating new project at `~/projects/ULG`

(ULG) pkg> add Plots

julia> using Plots

julia> plot([1; 2],[2; 3])

(ULG) pkg> activate
  Activating project at `~/.julia/environments/v1.8`

julia> using Plots
 │ Package Plots not found, but a package named Plots is available from a registry. 
 │ Install package?
 │   (@v1.8) pkg> add Plots 
 └ (y/n) [y]: 

(@v1.8) pkg> add Revise

(@v1.8) pkg> activate .

(ULG) pkg> add Revise@3.3.0

CC BY-SA 4.0 - Stephan Antholzer, Gregor Ehrensperger, Johannes Sappl. Last modified: August 31, 2023. Website built with Franklin.jl and the Julia programming language.