Software dependency graphs
A dependency graph shows the dependency relations between software
packages. Each node of the graph is a package, and an arrow from package
A to package B indicates that package A depends on package B. The
dependency graphs on this page have been obtained using the Guix
package manager, using the command
guix graph.
They include only dependencies that are specific to some piece of
software. Universal dependencies such as the operating system or the
package manager itself are not shown.
Since dependency graphs tend to become quite big, they are proposed here as PDF files, allowing you to zoom in for details.
There are several ways to define software dependencies, two of which will be illustrated here.
Run-time dependencies
The run-time dependencies of package A are all packages that need to be available on the same computer in order to be able to execute software from package A. Typical run-time dependencies are dynamic libraries that are required by an executable program, or interpreters for programs written in interpreted language. But any kind of software can be a run-time dependency. For a shell script, for example, all the programs called from the script are run-time dependencies.
The run-time dependency graphs shown in the following have been obtained using
guix graph -t references <package-1> <package-2> ...The graph type "references" shows dependencies at the file level. An arrow from package A to package B means that some file in package A depends on at least one file in package B. This explains that some arrows point from a package to itself, indicating internal dependencies.
Recursive build dependecies
Build dependencies are software packages required to convert other software from its source code into a form that can be executed on a computer. The most common build dependencies are compilers, but the category also includes pre-processors, code generators, and utilities such as configuration or build management tools. Recursive dependencies are the immediate dependencies plus the dependencies' dependencies. Recursive build dependencies are thus the most widely defined dependencies, including all software other than the operating system that is somehow involved in the construction or use of a piece of software. Formulated negatively: any software not in the recursive build dependencies of package A, other than the operating system, can not have any impact on results obtained by running software from package A.
The recursive build dependency graphs shown in the following have been obtained using
guix graph -t bag-emerged <package-1> <package-2> ...OpenBabel
OpenBabel is a data and file format converter for computational chemistry, written in the C language. I have included it here because it is a relatively simple program with few dependencies.
Run-time dependency graph
9 packages, 27 dependency relations
View as PDF
Recursive build dependency graph
121 packages, 2209 dependency relations
View as PDF
Scikit-learn
The library scikit-learn is a widely used machine learning library written in the Python language. It is included in this collection as a typical example of modern scientific software that combines computation and visualization functionality.
Run-time dependency graph
38 packages, 153 dependency relations
View as PDF
Recursive build dependency graph
485 packages, 10715 dependency relations
View as PDF
FEniCS
FEniCS is a computing platform for solving partial differential equations. It combines a computational code written in C++ with a Python interface for scripting and a collection of visualization tools. Scientific software packages of this size are rare, but they are also widely used.
Run-time dependency graph
112 packages, 587 dependency relations
View as PDF
Recursive build dependency graph
531 packages, 11841 dependency relations
View as PDF