Support for Managing Existing Changelog File

[bflad]

Description

Currently, changelog-build covers the case of building a CHANGELOG section for a current release. It would be great if there was the ability to manage an existing CHANGELOG file to add that section.

For example, the Terraform AWS Provider uses a script to build an updated CHANGELOG:

• Leave top header ("Unreleased"/"Next")
• changelog-build contents
• Append the previous release CHANGELOG contents

If this functionality would be acceptable for changelog-build, it would only need two additional pieces of optional information:

• CHANGELOG file name to manage
• Enable leaving a top header (e.g. skip line 1, skip line 1 if matching string, or add this line 1)

References

https://github.com/hashicorp/terraform-provider-aws/blob/ad8293cc2ec8556783e4896eed33ae97e56f5919/scripts/generate-changelog.sh#L1-L54

[bflad]

Having the top header handling be a flag to configure the line would remove any awkward manual step necessary after releases to put the new header there in the first place and repeatedly dance around it.

changelog-build -this-header "# Unreleased" -changelog-file CHANGELOG.md # ... other options

• Grab existing -changelog-file contents
• Overwrite -changelog-file as:
  • -this-header
  • Existing changelog-build functionality
  • Previous -changelog-file contents

[radeksimko]

This looks semi-related to #1

[david-yu]

We would love to see this be implemented on Consul, Consul K8s and Consul API Gateway, since this would tremendously help visualize what is coming in an upcoming release prior to the release happening and a full changelog generated.

[jmurret]

Our current Changelog has become so big that github often fails when trying to render it.

We have started discussing:

• Do we actually need this file anymore / should we just use the Github Release notes?
• If we kept it, do we change it to just contain links to the GH releases?

With these things only getting larger, I think our other products might run into similar issues that this is just too much data to present to anyone in a single view.