See the fragment files in the changelog.d directory.
- Fixed the documentation build on ReadTheDocs. Fixes issue 118.
- RST to Markdown conversion can now be stricter. Using the
--fail-if-warnoption on thescriv github-releasescommand will fail the command if your RST conversion generates warnings, for example due to malformed link references. - The
scriv github-releasecommand now has a--check-linksoption to check URLs. Each is fetched, and if an error occurs, warnings will show the URLs that didn't succeed.
- Commands no longer display full tracebacks for exceptions raised by scriv code.
- Literals can be extracted from .cabal files. Thanks Javier Sagredo.
- Use the git config
scriv.user_nickfor the user nick part of the fragment file. Thanks to Ronny Pfannschmidt, fixing issue 103. - Settings can now be prefixed with
command:to execute the rest of the setting as a shell command. The output of the command will be used as the value of the setting.
- If there are no changelog fragments,
scriv collectnow exits with status code of 2, fixing issue 110. - Changelogs with non-version headings now produce an understandable error
message from
scriv collect, thanks to James Gerity, fixing issue 100.
- The Version class introduced in 1.3.0 broke the
scriv github-releasecommand. This is now fixed.
.cfgfiles can now be read withliteral:settings, thanks to Matias Guijarro.
- In compliance with PEP 440, comparing version numbers now ignores a leading "v" character. This makes scriv more flexible about how you present version numbers in various places (code literals, changelog entries, git tags, and so on). Fixes issue 89.
- Scriv would fail trying to import tomllib on Python <3.11 if installed
without the
[toml]extra. This is now fixed, closing issue 80. - Settings specified as
file:will now search in the changelog directory and then the current directory for the file. The only exception is if the first component is.or.., then only the current directory is considered. Fixes issue 82. - Python variables with type annotations can now be read with
literal:settings, fixing issue 85. - Error messages for mis-formed
literal:configuration values are more precise, as requested in issue 84. - Error messages from settings validation are ScrivExceptions now, and report configuration problems more clearly and earlier in some cases.
scriv github-releasenow has a--repo=option to specify which GitHub repo to use when you have multiple remotes.
- Improved the error messages from
scriv github-releasewhen a GitHub repo can't be identified among the git remotes.
- The
scriv github-releasecommand has a new setting,ghrel_template. This is a template to use when building the release text, to add text before or after the Markdown extracted from the changelog. - The
scriv github-releasecommand now has a--dry-runoption to show what would happen, without posting to GitHub.
- File names specified for
file:settings will be interpreted relative to the current directory if they have path components. If the file name has no slashes or backslashes, then the old behavior remains: the file will be found in the fragment directory, or as a built-in template. - All exceptions raised by Scriv are now ScrivException.
- Parsing changelogs now take the insert-marker setting into account. Only content after the insert-marker line is parsed.
- More internal activities are logged, to help debug operations.
- Now literal configuration settings can be read from YAML files. Closes issue 69. Thanks, Florian Küpper.
- Fixed truncated help summaries by shortening them, closing issue 63.
- The
collectcommand now has a--title=TEXToption to provide the exact text to use as the title of the new changelog entry. Finishes issue 48.
The
github_releasecommand now only considers the top-most entry in the changelog. You can use the--alloption to continue the old behavior of making or updating GitHub releases for all of the entries.This change makes it easier for projects to start using scriv with an existing populated changelog file.
Closes issue 57.
- If there were no fragments to collect, scriv collect would make a new empty section in the changelog. This was wrong, and is now fixed. Now the changelog remains unchanged in this case. Closes issue 55.
- The
github-releasecommand will now issue a warning for changelog entries that have no version number. These can't be made into releases, so they are skipped. (issue 56).
scriv collectwill end with an error now if the version number would duplicate a version number on an existing changelog entry. Fixes issue 26.
- The
github_releasecommand will use a GitHub personal access token stored in the GITHUB_TOKEN environment variable, or from a .netrc file.
- The github_release command was using git tags as a command when it should have used git tag.
- Anchors in the changelog were being included in the previous sections when creating GitHub releases. This has been fixed, closing issue 53.
- Quoted commands failed, so we couldn't determine the GitHub remote.
- Added docs for
scriv github-release.
- Call pandoc properly on Windows for the github_release command.
- Dropped support for Python 3.6.
- The github-release command parses the changelog and creates GitHub releases from the entries. Changed entries will update the corresponding release.
- Added a
--versionoption.
- Parsing of fragments now only attends to the top-level section headers, and includes nested headers instead of splitting on all headers.
- Add an anchor before each version section in the output of
scriv collectso URLs for the sections are predictable and stable for each new version (Fixes issue 46). Thanks Abhilash Raj and Rodrigo Girão Serrão.
- Markdown fragments weren't combined properly. Now they are. Thanks Rodrigo Girão Serrão.
- Support finding version information in TOML files (like
pyproject.toml) using theliteralconfiguration directive. Thanks, Kurt McKee
- Fragment files in the fragment directory will be skipped if they match the
new configuration value
skip_fragments, a glob pattern. The default value is "README.*". This lets you put a README.md file in that directory to explain its purpose, as requested in issue 40.
- Switched from "toml" to "tomli" for reading TOML files.
- Setting
format=mddidn't properly cascade into other default settings, leaving you with RST settings that needed to be explicitly overridden (issue 39). This is now fixed.
- A new poorly documented API is available. See the Scriv, Changelog, and Fragment classes in the scriv.scriv module.
- Python 3.6 is now the minimum supported Python version.
- The changelog is now always written as UTF-8, regardless of the default encoding of the system. Thanks, Hei (yhlam).
- Settings can now be read from a pyproject.toml file. Install with the "[toml]" extra to be sure TOML support is available. Closes issue 9.
- Added the Philosophy section of the docs.
- The default entry header no longer puts the version number in square brackets: this was a misunderstanding of the keepachangelog formatting.
- Respect the existing newline style of changelog files. (#14) This means that a changelog file with Linux newlines on a Windows platform will be updated with Linux newlines, not rewritten with Windows newlines. Thanks, Kurt McKee.
- Support Windows' directory separator (
\) in unit test output. (#15) This allows the unit tests to run in Windows environments. Thanks, Kurt McKee. - Explicitly specify the directories and files that Black should scan. (#16) This prevents Black from scanning every file in a virtual environment. Thanks, Kurt McKee.
- Using "literal:" values in the configuration file didn't work on Python 3.6 or 3.7, as reported in issue 18. This is now fixed.
- Packaging fix.
- Markdown format is supported, both for fragments and changelog entries.
- Fragments can be mixed (some .rst and some .md). They will be collected and output in the format configured in the settings.
- Documentation.
- "python -m scriv" now works.
- The version number is displayed in the help message.
- When editing a new fragment during "scriv create", if the edited fragment has no content (only comments or blank lines), then the create operation will be aborted, and the file will be removed. (Closes issue 2.)
- If the fragment directory doesn't exist, a simple direct message is shown, rather than a misleading FileNotFound error (closes issue 1).
- When not using categories, comments in fragment files would be copied to the changelog file (issue 3). This is now fixed.
- RST syntax is better understood, so that hyperlink references and directives will be preserved. Previously, they were mistakenly interpreted as comments and discarded.
- Added the collect command.
- Configuration is now read from setup.cfg or tox.ini.
- A new configuration setting, rst_section_char, determines the character used in the underlines for the section headings in .rst files.
- The new_entry_template configuration setting is the name of the template
file to use when creating new entries. The file will be found in the
fragment_directory directory. The file name defaults to
new_entry.FMT.j2. If the file doesn't exist, an internal default will be used. - Now the collect command also includes a header for the entire entry. The underline is determined by the "rst_header_char" settings. The heading text is determined by the "header" setting, which defaults to the current date.
- The categories list in the config can be empty, meaning entries are not categorized.
- The create command now accepts --edit (to open the new entry in your text editor), and --add (to "git add" the new entry).
- The collect command now accepts --edit (to open the changelog file in an editor after the new entries have been collected) and --add (to git-add the changelog file and git rm the entries).
- The names of the main git branches are configurable as "main_branches" in the configuration file. The default is "master", "main", and "develop".
- Configuration values can now be read from files by prefixing them with "file:". File names will be interpreted relative to the changelog.d directory, or will be found in a few files installed with scriv.
- Configuration values can interpolate the currently configured format (rst or md) with "${config:format}".
- The default value for new templates is now "file: new_entry.${config:format}.j2".
- Configuration values can be read from string literals in Python code with a "literal:" prefix.
- "version" is now a configuration setting. This will be most useful when used with the "literal:" prefix.
- By default, the title of collected changelog entries includes the version if it's defined.
- The collect command now accepts a
--versionoption to set the version name used in the changelog entry title.
- RST now uses minuses instead of equals.
- The create command now includes the time as well as the date in the entry file name.
- The --delete option to collect is now called --keep, and defaults to False. By default, the collected entry files are removed.
- Created file names now include the seconds from the current time.
- "scriv create" will refuse to overwrite an existing entry file.
- Made terminology more uniform: files in changelog.d are "fragments." When collected together, they make one changelog "entry."
- The title text for the collected changelog entry is now created from the "entry_title_template" configuration setting. It's a Jinja2 template.
- Combined the rst_header_char and rst_section_char settings into one: rst_header_chars, which much be exactly two characters.
- Parsing RST fragments is more flexible: the sections can use any valid RST header characters for the underline. Previously, it had to match the configured RST header character.
- Fragments with no category header were being dropped if categories were in use. This is now fixed. Uncategorized fragments get sorted before any categorized fragments.
- Doesn't really do anything yet.