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

Hie @stevemao

Congratulations for the work you've done on this project. I have a question:

I follow the angular commit convention on my projects, but ... launching your changelog generator, I realized that I missed the column ...

• mine: feat(scope) subject
• should be: feat(scope): subject

Yeah, you can laugh :D ... All my commits are like that ...

So conventional-changelog-angular doesn't work out of the box. It simply needs a little adjustement here where the regexp :

• instead of /^(\w*)(?:\((.*)\))?\: (.*)$/
• would be /^(\w*)(?:\((.*)\))?\:? (.*)$/

It should be totally backwards compatible with the actual version but if I were you, as a maintainer, I wouldn't accept that kind of pull request because it's too much user specific, so I think the right way is that I make my own preset, based on yours.

Are you ok with that ? Just say yes, I'll manage everything myself ;D

Tophe

Add a bool prop to disable the Touchable container for the currently selected button.

This allows for a more accurate reproduction of SegmentedControlIOS highlighting during onPress and avoids possibly unnecessary reloads.

Pull Request to follow.

Conventional Changelog only work in github. It's possible support for GitLab?

I am having trouble getting commit messages such as (excerpt from my git log)

commit ad9f41ebe60c759bde22ed98b5caa0d24fa0793a
Author: cklein <[email protected]>
Date:   Sun May 1 09:47:25 2016 +0200

    refactor(tokens): comment and emptyline are simple tokens

commit 5cdc98136ac344c6d45887c39dd9a9b4f24a8a95
Author: cklein <[email protected]>
Date:   Fri Apr 29 20:55:08 2016 +0200

    fix(Authorship): authorship is whitespace

commit b2f70f7957ea60c7022262fbaf9132f4c91bf206
Author: cklein <[email protected]>
Date:   Fri Apr 29 20:53:20 2016 +0200

    fix(Comment): comment is whitespace

commit 0254948294647886e6bdd89ae4a7fb44ad1a0c2e
Author: cklein <[email protected]>
Date:   Tue Apr 26 19:48:02 2016 +0200

    version bump v0.0.10
    [ci skip]

to work with the new standard-changelog or when using conventional-changelog-cli and the angular preset.

Except for fix nothing else will make it into the changelog:

<a name="0.0.10"></a>
## [0.0.10](https://github.com/coldrye-es/ypo-parser-common/compare/v0.0.9...v0.0.10) (2016-05-01)

### Bug Fixes

* **Authorship:** authorship is whitespace ([5cdc981](https://github.com/coldrye-es/ypo-parser-common/commit/5cdc981))
* **Comment:** comment is whitespace ([b2f70f7](https://github.com/coldrye-es/ypo-parser-common/commit/b2f70f7))

<a name="0.0.9"></a>

Am I doing something wrong?

There are numerous conditionals in the writerOpts transform function to map the origin commit type to a new name.

var typeMappings = {
  feat: 'Features'
}

commit.type = typeMappings[commit.type] || commit.type;

It would help keep cyclomatic complexity down. (jshint is warning on that long list of if statements)

I'm in the process of extending conventional-changelog-angular in the form of a fork, but couldn't figure out why typeof commit.hash === 'string' was necessary. In what cases would the hash not be a string?

If it's needed could a unit test be added to help get the project up to 100% test coverage.

Any special reason to ignore them? The angular commit convention doesn't stipulate anything about it.

Hi,

I've been using the Angular commit messages convention fro more than a year now, and I find it great.

On the other hand, I think there's one type of commits that is very relevant, and it's missing from the list: the "content" type.

Every time I'm working on an app or a website and I need to commit changes to its static content (e.g. text in a page, pictures, titles, etc.) I can't find a good fit in the existing commit types included in the list.

In fact, it's neither "feat", nor "style", nor "docs". Apart from this, I've found the list to be pretty exhaustive.

Therefore, I propose to include the "content" type in the list, with the description "A content change that only affects static text or images" (or something similar).

Let me know your thoughts, and if you'd like me to create a PR for it!

Thanks, cheers :)

p.s. relates to commitizen/cz-conventional-changelog#15

In older version of conventional-changelog (0.0.17), changelog was automatically grouped not only by "Bug Fixes"/"Features" but also subgrouped by "scope". Like in example below:

As we can see "cart" scope become group for all commits from that scope. Is this possible to have same functionality in current version of angular preset? If no, how to achieve it?

I would like to overwrite the gitRawCommitsOpts.

conventionalChangelog(
      {
        preset: 'angular', releaseCount: 0,  gitRawCommitsOpts: {
          merges: true,
          noMerges: false
        }
      })

But it seems like that when using angular mode gitRawCommitsOpts is not used in the source code?

Please check:
https://github.com/conventional-changelog/conventional-changelog-angular/blob/master/index.js#L117

`conventional-changelog -p angular -i CHANGELOG.md -o CHANGELOG.md -s

will output:

<a name="2.0.13"></a>
## [2.0.13](/http://gitlab.alibaba-inc.com/wing/wing-node/compare/v2.0.12...v2.0.13) (2016-04-01)

i check the angular template, @root.host is empty, and / that immediate after @root.host is unnecessary.

<a name="{{version}}"></a>
{{#if isPatch}}##{{else}}#{{/if}} {{#if @root.linkCompare}}[{{version}}]({{@root.host}}/{{#if @root.owner}}

I'm in the process of extending conventional-changelog-angular in the form of a fork, but couldn't figure out why typeof commit.subject === 'string' was necessary.

If it's needed could a unit test be added to help get the project up to 100% test coverage.

The long chain of commit type conditionals doesn't have a test for the else case where no type matches.

In my fork of conventional-changelog-angular I added gitDummyCommit(['what(ngOptions): unknown type', 'BREAKING CHANGE: The Change is huge.']); to the BREAKING CHANGE test case, along with expect(chunk).to.include('what'); as an assertion.

That resolve that missing test case.