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

See this commit message: angular/angular@098b461

Notice it says "BREAKING CHANGES:" with an S at the end.

If I generate a changelog with preset: angular, releaseCount: 1 including that commit, the breaking change ends up having an "S" after the scope, like so:

* core: S:

Previously multiple template bindings on one element
(ex. `<div *ngIf='..' *ngFor='...'>`) were allowed but most of the time
were leading to undesired result. It is possible that a small number
of applications will see template parse errors that shuld be fixed by
nesting elements or using `<template>` tags explicitly.

A lot of commit messages add the S at the end of breaking changes, so I think it would be worthwhile to add a special case for that. Happy to open a PR if this sounds reasonable :).

Dependency versions:

• gulp-conventional-changelog 1.1.0
• conventional-changelog 1.1.0
• conventional-changelog-core 1.5.0

A very very minor bug

I recently updated gulp-conventional-changelog from 0.5.1 to 1.1.0. I use it with the angular preset.

When I update my repository from 1.0.0-alpha.7 to 1.0.0-alpha.8 the 'GitHub compare' link on the title will unnecessarily prefix the most recent tag with a v:
SteveVanOpstal/LegendBuilder@1.0.0-alpha.7...v1.0.0-alpha.8

For the link to work, it should be:
SteveVanOpstal/LegendBuilder@1.0.0-alpha.7...1.0.0-alpha.8

It seems like you assume tags are always prefixed with a v in merge-config.js

conventional-changelog/standard-version#33 😄

Hi, great library guys.

I started using this library for my project. While all the commits appear as output, they are not being grouped by type, which is meant to be the default for the writer?

I am running this as: node changelog.js

var conventionalChangelogCore = require('conventional-changelog-core');

conventionalChangelogCore({
    transform : function(commit, cb) {

      if(/^FIX/.test(commit.header)) {
        commit.type = 'bug';
      }

      if(/^NEW|FEATURE/.test(commit.header)) {
        commit.type = 'feat';
      }

      if(/^bump version --skip-ci/.test(commit.header)) {
        commit = null;
      }

      cb(null, commit);
    },
  })
  .pipe(process.stdout); // or any writable stream

Since version 1.3.3 is released our changelog task does not generate any commit links anymore.

My guess is that the change in lib/merge-config.js in commit eaa3b6f breaks it when calling the changelog task without a context object (and within a git repo).

repo.browse() will never be called because context.host is filled in through calling getPkgRepo(pkg) which calls git-remote-origin-url

1. ATM one have to depend on another. This means we can't do mocha --grep and only test certain tests so its hard to work with break points. Maybe we could detect if the git history is good before running the it block.
2. The construction of git history is partially in it blocks. This is fine but the performance tests with mocha would count those. Although we don't care about the milliseconds, this could be improved.

Low priority 😄

default context.linkCompare should also depend on context.repository or context.repoUrl

I use a package similar to conventional-changelog-angular with a single major difference.

Inside the parserOpts object (https://github.com/conventional-changelog/conventional-changelog-angular/blob/master/index.js#L18) I set issuePrefixes to ['#', '@']. However, only the # prefix is respected by conventional-commits-parser.

It seems this line is replacing my issuePrefixes value with the default for my SCM system (as retrieved from here).

If I comment out that line, my prefix is respected, and my changelog is generated correctly.

I have the following error when running [email protected]:

$ flow
node_modules/conventional-changelog-core/test/fixtures/_malformation.json:1
1:
^ Unexpected end of input

I fixed it by adding the following lines in my .flowconfig file:

[ignore]
.*/node_modules/conventional-changelog-core/.*\.json$

Would it makes sense to remove the tests files from the package?