Change log

This change log follows the Keep a Changelog spec. Every release contains the following sections:

  • Added for new features.

  • Changed for changes in existing functionality.

  • Deprecated for soon-to-be removed features.

  • Removed for now removed features.

  • Fixed for any bug fixes.

  • Security in case of vulnerabilities.

The versions follow semantic versioning.

1.0.0 - 2022-05-19

A major release! Do not worry, no breaking changes but a development team (@carmenbianca, @floriansnow, @linozen, @mxmehl and @nicorikken) that is confident enough to declare the REUSE helper tool stable, and a bunch of long-awaited features!

Apart from smaller changes under the hood and typical maintenance tasks, the main additions are new flags to the addheader subcommand that ease recursive and automatic operations, the ability to ignore areas of a file that contain strings that may falsely be detected as copyright or license statements, and the option to merge copyright lines. The tool now also has better handling of some edge cases with copyright and license identifiers.

We would like to thank the many contributors to this release, among them @ajinkyapatil8190, @aspiers, @ferdnyc, @Gri-ffin, @hexagonrecursion, @hoijui, @Jakelyst, @Liambeguin, @rex4539, @robinkrahl, @rpavlik, @siiptuo, @thbde and @ventosus.


  • Extend tool documentation with scripts to help using this tool and automating some steps that are not built into the tool itself. (#500)

  • Recommendations for installation/run methods: package managers and pipx (#457)

  • Docker images for AArch64 (#478)

  • Added the ability to ignore parts of a file when running reuse lint. Simply add REUSE-IgnoreStart and REUSE-IgnoreEnd as comments and all lines between the two will be ignored by the next run of reuse lint. (#463)

  • Meson subprojects are now ignored by default. (#496)

  • More file types are recognised:

    • sbt build files (.sbt)

    • Vimscript files (.vim)

  • Added --skip-existing flag to addheader in order to skip files that already contain SPDX information. This may be useful for only adding SPDX information to newly created files. (#480)

  • Added --recursive flag to addheader. (#469)

  • Preserve shebang for more script files:

    • V-Lang (#432)

  • Ignore all SPDX files with their typical formats and extensions. (#494)

  • Add support for merging copyright lines based on copyright statement, transforming multiple lines with a single year into a single line with a range. (#328)


  • Use setuptools instead of the deprecated distutils which will be removed with Python 3.12. (#451)

  • addheader --explicit-license renamed to --force-dot-license. (#476)

  • Dockerfiles for reuse-tool are now in a separate subdirectory docker. (#499)

  • Updated SPDX license list to 3.17. (#513)

  • The copyright detection mechanism now silently accepts the following strings: Copyright(c) and Copyright(C). (#440)


  • Deprecated --explicit-license in favour of --force-dot-license. --explicit-license will remain useable (although undocumented) for the foreseeable future. (#476)


  • JsxCommentStyle in favor of using CCommentStyle directly (see section Fixed). (#406)


  • Better support for unary “+” operator in license identifiers. For example, if Apache-1.0+ appears as a declared license, it should not be identified as missing, bad, or unused if LICENSES/Apache-1.0.txt exists. It is, however, identified separately as a used license. (#123)

  • When addheader creates a .license file, that file now has a newline at the end. (#477)

  • Cleaned up internal string manipulation. (#477)

  • JSX (.jxs and .tsx) actually uses C comment syntax as JSX blocks never stand at the beginning of the file where the licensing info needs to go. (#406)

0.14.0 - 2021-12-27

Happy holidays! This is mainly a maintenance release fixing some subcommands and adding loads of supported file types and file names. However, you can also enjoy the supported-licenses subcommand and the --quiet flag for linting as well as better suggestions for license identifiers. Thanks to everyone who contributed!


  • supported-licenses command that lists all licenses supported by REUSE (#401)

  • --quiet switch to the lint command (#402)

  • Better suggestions for faulty SPDX license identifiers in download and init (#416)

  • Python 3.10 support declared

  • More file types are recognised:

    • Apache FreeMarker Template Language (.ftl)

    • AsciiDoc (.adoc, .asc, .asciidoc)

    • Bibliography (.csl)

    • C++ (.cc and .hh)

    • GraphQL (.graphql)

    • Handlebars (.hbs)

    • Markdown-linter config (.mdlrc)

    • MS Office (.doc, .xls, .pptx and many more)

    • Nimble (.nim.cfg, .nimble)

    • Open Document Format (.odt, .ods, .fodp and many more)

    • Perl plain old documentation (.pod)

    • Portable document format (.pdf)

    • Protobuf files (.proto)

    • Soy templates (.soy)

    • SuperCollider (.sc, .scsyndef)

    • Turtle/RDF (.ttl)

    • V-Lang (.v, .vsh)

    • Vue.js (.vue)

  • More file names are recognised:

    • Doxygen (Doxyfile)

    • ESLint (.eslintignore and .eslintrc)

    • Meson options file (meson_options.txt)

    • NPM ignore (.npmignore)

    • Podman container files (Containerfile)

    • SuperCollider (archive.sctxar)

    • Yarn package manager (.yarn.lock and .yarnrc)


  • Updated SPDX license list to 3.15


  • Fix Extensible Stylesheet Language (.xsl) to use HTML comment syntax

  • Allow creating .license file for write-protected files (#347) (#418)

  • Do not break XML files special first line (#378)

  • Make download subcommand work correctly outside of project root and with --root (#430)

0.13.0 - 2021-06-11


  • addheader recognises file types that specifically require .license files instead of headers using UncommentableCommentStyle. (#189)

  • .hgtags is ignored. (#227)

  • spdx-symbol added to possible copyright styles. (#350)

  • addheader ignores case when matching file extensions and names. (#359)

  • Provide latest-debian as Docker Hub tag, created by Dockerfile-debian. (#321)

  • More file types are recognised:

    • Javascript modules (.mjs)

    • Jupyter Notebook (.ipynb)

    • Scalable Vector Graphics (.svg)

    • JSON (.json)

    • Comma-separated values (.csv)

    • Racket (.rkt)

    • Org-mode (.org)

    • LaTeX package files (.sty)

    • devicetree (.dts, .dtsi)

    • Bitbake (.bb, .bbappend, .bbclass)

    • XML schemas (.xsd)

    • OpenSCAD (.scad)

  • More file names are recognised:

    • Bash configuration (.bashrc)

    • (.coveragerc)

    • Jenkins (Jenkinsfile)

    • SonarScanner (

    • Gradle (, gradlew)


  • Bump alpine Docker base image to 3.13. (#369)


  • Fixed a regression where unused licenses were not at all detected. (#285)

  • Declared dependency on python-debian != 0.1.39 on Windows. This version does not import on Windows. (#310)

  • is now recognised instead of the incorrect by addheader. (#306)

  • addheader now checks whether a file is both readable and writeable instead of only writeable. (#241)

  • addheader now preserves line endings. (#308)

  • download does no longer fail when both --output and --all are used. (#326)

  • Catch erroneous SPDX expressions. (#331)

  • Updated SPDX license list to 3.13.

0.12.1 - 2020-12-17


  • Bumped versions of requirements. (#288)

0.12.0 - 2020-12-16

This release was delayed due to the absence of the lead developer (Carmen, me, the person writing these release notes). Many thanks to Max Mehl for coordinating the project in my absence. Many thanks also to the contributors who sent in pull requests, in reverse chronological order: Olaf Meeuwissen, Mikko Piuola, Wolfgang Traylor, Paul Spooren, Robert Cohn, ethulhu, pukkamustard, and Diego Elio Pettenò.


  • Separate Docker image with additional executables installed (fsfe/reuse:latest-extra) (#238)

  • Allow different styles of copyright lines: SPDX (default), String, String (C), String ©, and © (#248)

  • Convenience function to update resources (SPDX license list and exceptions) (#268)

  • More file types are recognised:

    • ClojureScript (.cljc, .cljs)

    • Fortran (.F, .F90, .f90, .f95, .f03, .f, .for)

    • Makefile (.mk)

    • PlantUML (.iuml, .plantuml, .pu, .puml)

    • R (.R, .Renviron, .Rprofile)

    • ReStructured Text (.rst)

    • RMarkdown (.Rmd)

    • Scheme (.scm)

    • TypeScript (.ts)

    • TypeScript JSX (.tsx)

    • Windows Batch (.bat)

  • More file names are recognised:

    • .dockerignore

    • Gemfile

    • go.mod


    • Rakefile


  • Use UTF-8 explicitly when reading files (#242)


  • Updated license list to 3.11.

0.11.1 - 2020-06-08


  • Similar to CAL-1.0 and CAL-1.0-Combined-Work-Exception, SHL-2.1 is now ignored because it contains an SPDX tag within itself.

0.11.0 - 2020-05-25


  • Added --skip-unrecognised flag to addheader in order to skip files with unrecognised comment styles instead of aborting without processing any file.


  • Always write the output files encoded in UTF-8, explicitly. This is already the default on most Unix systems, but it was not on Windows.

  • All symlinks and 0-sized files in projects are now ignored.


  • The licenses CAL-1.0 and CAL-1.0-Combined-Work-Exception contain an SPDX tag within themselves. Files that are named after these licenses are now ignored.

  • Fixed a bug where addheader wouldn’t properly apply the template on .license files if the .license file was non-empty, but did not contain valid SPDX tags.

0.10.1 - 2020-05-14


  • Updated license list to 3.8-106-g4cfec76.

0.10.0 - 2020-04-24


  • Add support for autoconf comment style (listed as m4).

  • More file types are recognised:

    • Cython (.pyx, .pxd)

    • Sass and SCSS (.sass, .scss)

    • XSL (.xsl)

    • Mailmap (.mailmap)

  • Added --single-line and --multi-line flags to addheader. These flags force a certain comment style.


  • The Docker image has an entrypoint now. In effect, this means running:

    docker run -v $(pwd):/data fsfe/reuse lint

    instead of

    docker run -v $(pwd):/data fsfe/reuse reuse lint.

0.9.0 - 2020-04-21


  • Added support for Mercurial 4.3+.

  • A pre-commit hook has been added.

  • When an incorrect SPDX identifier is forwarded to download or init, the tool now suggests what you might have meant.


  • Under the hood, a lot of code that has to do with Git and Mercurial was moved into its own module.

  • The Docker image has been changed such that it now automagically runs reuse lint on the /data directory unless something else is specified by the user.


  • Fixed a bug with addheader --explicit-license that would result in file.license.license if file.license already existed.

  • Fixed a Windows-only bug to do with calling subprocesses.

  • Fixed a rare bug that would trigger when a directory is both ignored and contains a .git file.

0.8.1 - 2020-02-22


  • Support Jinja (Jinja2) comment style.

  • Support all multi-line comment endings when parsing for SPDX information.


  • Improvements to German translation by Thomas Doczkal.

  • No longer remove newlines at the end of files when using addheader.

  • There can now be a tab as whitespace after SPDX-License-Identifier and SPDX-FileCopyrightText.

0.8.0 - 2020-01-20


  • Implemented --root argument to specify the root of the project without heuristics.

  • The linter will complain about licenses without file extensions.

  • Deprecated licenses are now recognised. lint will complain about deprecated licenses.

  • ProjectReport generation (lint, spdx) now uses Python multiprocessing, more commonly called multi-threading outside of Python. This has a significant speedup of approximately 300% in testing. Because of overhead, performance increase is not exactly linear.

  • For setups where multiprocessing is unsupported or unwanted, --no-multiprocessing is added as flag.

  • addheader now recognises many more extensions. Too many to list here.

  • addheader now also recognises full filenames such as Makefile and .gitignore.

  • Added BibTex comment style.

  • Updated translations:

    • Dutch (André Ockers, Carmen Bianca Bakker)

    • French (OliBug, Vincent Lequertier)

    • Galician (pd)

    • German (Max Mehl)

    • Esperanto (Carmen Bianca Bakker)

    • Portuguese (José Vieira)

    • Spanish (Roberto Bauglir)

    • Turkish (T. E. Kalayci)


  • The linter output has been very slightly re-ordered to be more internally consistent.

  • reuse --version now prints a version with a Git hash on development versions. Towards that end, the tool now depends on setuptools-scm during setup. It is not a runtime dependency.


  • lint no longer accepts path arguments. Where previously one could do reuse lint SUBDIRECTORY, this is no longer possible. When linting, you must always lint the entire project. To change the project’s root, use --root.

  • FileReportInfo has been removed. FileReport is used instead.


  • A license that does not have a file extension, but whose full name is a valid SPDX License Identifier, is now correctly identified as such. The linter will complain about them, however.

  • If the linter detects a license as being a bad license, that license can now also be detected as being missing.

  • Performance of project.all_files() has been improved by quite a lot.

  • Files with CRLF line endings are now better supported.

0.7.0 - 2019-11-28


  • The program’s package name on PyPI has been changed from fsfe-reuse to reuse. fsfe-reuse==1.0.0 has been created as an alias that depends on reuse. fsfe-reuse will not receive any more updates, but will still host the old versions.

  • For users of fsfe-reuse, this means:

    • If you depend on fsfe-reuse or fsfe-reuse>=0.X.Y in your requirements.txt, you will get the latest version of reuse when you install fsfe-reuse. You may like to change the name to reuse explicitly, but this is not strictly necessary.

    • If you depend on fsfe-reuse==0.X.Y, then you will keep getting that version. When you bump the version you depend on, you will need to change the name to reuse.

    • If you depend on fsfe-reuse>=0.X.Y<1.0.0, then 0.6.0 will be the latest version you receive. In order to get a later version, you will need to change the name to reuse.

0.6.0 - 2019-11-19


  • --include-submodules is added to also include submodules when linting et cetera.

  • addheader now also recognises the following extensions:

    • .kt

    • .xml

    • .yaml

    • .yml


  • Made the workaround for MachineReadableFormatError introduced in 0.5.2 more generic.

  • Improved shebang detection in addheader.

  • For addheader, the SPDX comment block now need not be the first thing in the file. It will find the SPDX comment block and deal with it in-place.

  • Git submodules are now ignored by default.

  • addheader --explicit-license now no longer breaks on unsupported filetypes.

0.5.2 - 2019-10-27


  • python3 -m reuse now works.


  • Updated license list to 3.6-2-g2a14810.


  • Performance of reuse lint improved by at least a factor of 2. It no longer does any checksums on files behind the scenes.

  • Also handle MachineReadableFormatError when parsing DEP5 files. Tries to import that error. If the import is unsuccessful, it is handled.

0.5.1 - 2019-10-24 [YANKED]

This release was replaced by 0.5.2 due to importing MachineReadableFormatError, which is not a backwards-compatible change.

0.5.0 - 2019-08-29


  • TeX and ML comment styles added.

  • Added --year and --exclude-year to reuse addheader.

  • Added --template to reuse addheader.

  • Added --explicit-license to reuse addheader.

  • binaryornot added as new dependency.

  • Greatly improved the usage documentation.


  • reuse addheader now automatically adds the current year to the copyright notice.

  • reuse addheader preserves the original header below the new header if it did not contain any SPDX information.

  • reuse addheader now correctly handles .license files.

  • Bad licenses are no longer resolved to LicenseRef-Unknown. They are instead resolved to the stem of the path. This reduces the magic in the code base.

  • .gitkeep files are now ignored by the tool.

  • Changed Lisp’s comment character from ‘;;’ to ‘;’.

0.4.1 - 2019-08-07


  • --all argument help to reuse download, which downloads all detected missing licenses.


  • When using reuse addheader on a file that contains a shebang, the shebang is preserved.

  • Copyright lines in reuse spdx are now sorted.

  • Some publicly visible TODOs were patched away.

0.4.0 - 2019-08-07

This release is a major overhaul and refactoring of the tool. Its primary focus is improved usability and speed, as well as adhering to version 3.0 of the REUSE Specification.


  • reuse addheader has been added as a way to automatically add copyright statements and license identifiers to the headers of files. It is currently not complete.

  • reuse init has been added as a way to initialise a REUSE project. Its functionality is currently scarce, but should improve in the future.


  • reuse lint now provides a helpful summary instead of merely spitting out non-compliant files.

  • reuse compile is now reuse spdx.

  • In addition to Copyright and ©, copyright lines can be marked with the tag SPDX-FileCopyrightText:. This is the new recommended default.

  • Project no longer depends on pygit2.

  • The list of SPDX licenses has been updated.

  • Valid-License-Identifier is no longer used, and licenses and exceptions can now only live inside of the LICENSES/ directory.


  • Removed --ignore-debian.

  • Removed --spdx-mandatory, --copyright-mandatory, --ignore-missing arguments from reuse lint.

  • Remove reuse license.

  • GPL-3.0 and GPL-3.0+ (and all other similar GPL licenses) are no longer detected as SPDX identifiers. Use GPL-3.0-only and GPL-3.0-or-later instead.


  • Scanning a Git directory is a lot faster now.

  • Scanning binary files is a lot faster now.

0.3.4 - 2019-04-15

This release should be a short-lived one. A new (slightly backwards-incompatible) version is in the works.


  • Copyrights can now start with © in addition to Copyright. The former is now recommended, but they are functionally similar.


0.3.3 - 2018-07-15


  • Any files with the suffix .spdx are no longer considered licenses.

0.3.2 - 2018-07-15


  • The documentation now builds under Python 3.7.

0.3.1 - 2018-07-14


  • When using reuse from a child directory using pygit2, correctly find the root.

0.3.0 - 2018-05-16


  • The output of reuse compile is now deterministic. The files, copyright lines and SPDX expressions are sorted alphabetically.


  • When a GPL license could not be found, the correct -only or -or-later extension is now used in the warning message, rather than a bare GPL-3.0.

  • If you have a license listed as SPDX-Valid-License: GPL-3.0-or-later, this now correctly matches corresponding SPDX identifiers. Still it is recommended to use SPDX-Valid-License: GPL-3.0 instead.

0.2.0 - 2018-04-17


  • Internationalisation support added. Initial support for:

    • English.

    • Dutch.

    • Esperanto.

    • Spanish.


  • The license list of SPDX 3.0 has deprecated GPL-3.0 and GPL-3.0+ et al in favour of GPL-3.0-only and GPL-3.0-or-later. The program has been amended to accommodate sufficiently for those licenses.


  • Project.reuse_info_of now extracts, combines and returns information both from the file itself and from debian/copyright.

  • ReuseInfo now holds sets instead of lists.

    • As a result of this, ReuseInfo will not hold duplicates of copyright lines or SPDX expressions.

  • click removed as dependency. Good old argparse from the library is used instead.

0.1.1 - 2017-12-14


  • The reuse --help text has been tidied up a little bit.


  • Release date in change log fixed.

  • The PyPI homepage now gets reStructuredText instead of Markdown.

0.1.0 - 2017-12-14


  • Successfully parse old-style C and HTML comments now.

  • Added reuse compile, which creates an SPDX bill of materials.

  • Added --ignore-missing to reuse lint.

  • Allow to specify multiple paths to reuse lint.

  • chardet added as dependency.

  • pygit2 added as soft dependency. reuse remains usable without it, but the performance with pygit2 is significantly better. Because pygit2 has a non-Python dependency (libgit2), it must be installed independently by the user. In the future, when reuse is packaged natively, this will not be an issue.


  • Updated to version 2.0 of the REUSE recommendations. The most important change is that License-Filename is no longer used. Instead, the filename is deducted from SPDX-License-Identifier. This change is NOT backwards compatible.

  • The conditions for linting have changed. A file is now non-compliant when:

    • The license associated with the file could not be found.

    • There is no SPDX expression associated with the file.

    • There is no copyright notice associated with the file.

  • Only read the first 4 KiB (by default) from code files rather than the entire file when searching for SPDX tags. This speeds up the tool a bit.

  • Project.reuse_info_of no longer raises an exception. Instead, it returns an empty ReuseInfo object when no reuse information is found.

  • Logging is a lot prettier now. Only output entries from the reuse module.


  • reuse --ignore-debian compile now works as expected.

  • The tool no longer breaks when reading a file that has a non-UTF-8 encoding. Instead, chardet is used to detect the encoding before reading the file. If a file still has errors during decoding, those errors are silently ignored and replaced.

0.0.4 - 2017-11-06


  • Removed dependency on os.PathLike so that Python 3.5 is actually supported

0.0.3 - 2017-11-06


  • Fixed the link to PyPI in the README.

0.0.2 - 2017-11-03

This is a very early development release aimed at distributing the program as soon as possible. Because this is the first release, the changelog is a little empty beyond “created the program”.

The program can do roughly the following:

  • Detect the license of a given file through one of three methods (in order of precedence):

    • Information embedded in the .license file.

    • Information embedded in its header.

    • Information from the global debian/copyright file.

  • Find and report all files in a project tree of which the license could not be found.

  • Ignore files ignored by Git.

  • Do some logging into STDERR.