conventional-changelog-archived-repos / conventional-changelog-cli Goto Github PK

Latest version of changelog-cli wants an exported object

if (flags.config) {
    config = require(resolve(process.cwd(), flags.config));
  } else {
    config = {};
  }
// [...]

var changelogStream = conventionalChangelog(options, templateContext, config.gitRawCommitsOpts, config.parserOpts, config.writerOpts)

while latest config version of conventional-changelog-angular exports a promise and other versions a callback.

That's why the conventional-changelog library works with these configuration files but not the CLI.

In the following workflow people could combine 4 to 7 with a simple hook.
https://github.com/conventional-changelog-archived-repos/conventional-changelog-cli#recommended-workflow

With this hook people could simply use npm version

{
  "scripts": {
    "version": "conventional-changelog -p angular -i CHANGELOG.md -s -r 0 && git add CHANGELOG.md"
  }
}

If you combine with a proper .npmrcthe workflow could become:

1. Make changes
2. Commit those changes
3. Make sure Travis turns green
4. Bump version with npm version <patch|minor|major> command
5. Push

I find it strange for a rather "simple" tool to change its general behavior between minor releases, e.g. switching from -w to -s and now requiring -p <preset> in order to be able to get an expected result when using conventional-changelog.

While I appreciate you all doing a great job in providing this tool, it would be nice of you for at least keep a reasonable default when breaking with existing behavior.

The most reasonable defaults would have been

• aliasing -w with -s
• using a default preset -p angular

TIA

Currently is append on the bottom of the changelog file :(

First I do:

conventional-changelog --preset=angular -infile=CHANGELOG.md --same-file --release-count=0

later I release a new tag and:

conventional-changelog --preset=angular -infile=CHANGELOG.md -a -s

hmm... Totally missed that.