Implementation:Mlflow Mlflow Update Changelog

Knowledge Sources	Mlflow_Mlflow
Domains	Release Automation, Documentation
Last Updated	2026-02-13 20:00 GMT

Overview

A development script that automatically generates the CHANGELOG.md entry for a new MLflow release by extracting pull request metadata from git history and the GitHub GraphQL API.

Description

update_changelog.py is part of the MLflow release automation workflow. It performs the following steps:

1. Determines the commits between the previous release tag and the current release branch using git log with cherry-pick filtering.
2. Extracts PR numbers from commit messages matching the pattern (#NNNN).
3. Batch-fetches PR metadata (title, author, labels) using the GitHub GraphQL API, processing PRs in chunks of 50 to stay within query size limits.
4. Categorizes PRs by their rn/ (release note) labels into sections: breaking changes, highlights, features, bug fixes, documentation updates, and small updates.
5. Generates a formatted changelog entry with a version header, date, and categorized sections, then prepends it to CHANGELOG.md.

The script uses two key data classes: PullRequest (a NamedTuple) which provides formatted string representations including area labels and PR links, and Section (a NamedTuple) which formats categorized PR lists. The script validates that all PRs have release note labels and that no unknown labels are present, failing with an assertion error if categorization is incomplete.

Usage

Use this script when preparing a new MLflow release to generate the changelog entry. Requires the GH_TOKEN environment variable for GitHub API authentication.

Code Reference

Source Location

• Repository: Mlflow_Mlflow
• File: dev/update_changelog.py
• Lines: 1-281

Signature

class PullRequest(NamedTuple):
    title: str
    number: int
    author: str
    labels: list[str]

    @property
    def url(self) -> str: ...
    @property
    def release_note_labels(self) -> list[str]: ...

class Section(NamedTuple):
    title: str
    items: list[Any]

def get_header_for_version(version: str) -> str: ...
def extract_pr_num_from_git_log_entry(git_log_entry: str) -> int | None: ...
def format_label(label: str) -> str: ...
def is_shallow() -> bool: ...
def batch_fetch_prs_graphql(pr_numbers: list[int]) -> list[PullRequest]: ...
def main(prev_version: str, release_version: str, remote: str) -> None: ...

Import

# Run as a standalone script
python dev/update_changelog.py --prev-version 2.18.0 --release-version 2.19.0

I/O Contract

Inputs

Name	Type	Required	Description
--prev-version	str	Yes	The previous MLflow release version (e.g., "2.18.0")
--release-version	str	Yes	The MLflow version being released (e.g., "2.19.0")
--remote	str	No	Git remote to use (default: "origin")

Outputs

Name	Type	Description
CHANGELOG.md	file	Updated changelog file with new version entry prepended
stdout	text	Progress messages including batch fetch status

Usage Examples

Basic Usage

# Generate changelog for version 2.19.0
export GH_TOKEN="your-github-token"
python dev/update_changelog.py --prev-version 2.18.0 --release-version 2.19.0

Custom Remote

# Use a different git remote
python dev/update_changelog.py --prev-version 2.18.0 --release-version 2.19.0 --remote upstream

Release Note Label Mapping

The script recognizes the following rn/ labels for categorizing pull requests:

Label	Section
rn/breaking-change	Breaking changes
rn/highlight	Major new features
rn/feature	Features
rn/bug-fix	Bug fixes
rn/documentation	Documentation updates
rn/none	Small bug fixes and documentation updates (grouped by author)

PRs without any rn/ label cause the script to fail with an assertion error listing the uncategorized PR URLs.

Related Pages

• Environment:Mlflow_Mlflow_Python_Runtime_Environment