…t of the NPM box

…ck 'em all.

…t of the NPM box

…kages: via the `Npm-Scope` header. See also https://docs.npmjs.com/misc/registry

…tion of updates.

…kages: via the `Npm-Scope` header. See also https://docs.npmjs.com/misc/registry

…onsequently from the dev deps)

…nd augmenting github issues with an URL.

… output with `-c` on platforms/terminals which are not recognized as color-supporting. (e.g. Windows+msys bash)

# Conflicts:
#	package-lock.json

…older) due to `const` usage in the code. Discarding antiquated NodeJS support in `changelog` itself now: NodeJS versions 4 and later are supported and CI tested (adjusted TravisCI config accordingly)

…/6/7/8

…eport anymore: `expect.an.object` is not the correct use: http://chaijs.com/api/bdd/ --> `expect.an('object')`

…uild -2 and before, after having updated to chai 4)

…eport anymore: `expect.an.object` is not the correct use: http://chaijs.com/api/bdd/ --> `expect.an('object')`

# Conflicts:
#	package-lock.json

…m -rf node_modules; rm -f package-lock.json; npm i`

…pdated since 2013) to chalk-supports module (which is linked from rmp website's has-color entry). Added note to README for Windows users who run into issues around console TTY detection failing on several msys/mingw systems: `-c` vs. `--color`.

…y like console.log and console.error so we can log more complex (debug) messages easily.

…ned' then. Better validation in the code.

…e, i.e. load the project name from the local package.json file, also pick up the repository URL from that file as well and apply that as the actual github URL rather than going to the NPM registry and collecting the github repo URL from the data produced by *that* one: the reason we do this is that we MAY VERY PROBABLY have adjusted the package.json file to be a scoped package or otherwise, but have not yet changed the repository URL to point to our fork/clone/derivative: it's one of those fields you only correct when you find out they matter... and now it turns out they matter for the CHANGELOG tool but then, of course, you otherwise would have to push a new NPM release just to correct this, which is ludicrous, so we have the local package.json take priority when we obtain data from it anyway. (Ergo: specifying the npm repo identifier on the command line WILL NOT take the package.json repository URL as explicit specification of the npm registry entry on the command line results in CHANGELOG not accessing any local package.json *at all*!)

…ntly reverted to old behaviour. See also commit SHA-1: 368495e :: fix: npm.org now has a different way to request info about scoped packages: via the `Npm-Scope` header. See also https://docs.npmjs.com/misc/registry

…ntly reverted to old behaviour. See also commit SHA-1: 368495e :: fix: npm.org now has a different way to request info about scoped packages: via the `Npm-Scope` header. See also https://docs.npmjs.com/misc/registry

….json

…le `assert.isObject()` does what it's supposed to do.

… are in sync with npm releases, at least when they have a 'semver' format, so we collect those alongside the ones produced by the npm registry.

This is particularly handy to obtain a complete releases set (CLI: 'all') when a project is worked on via fork + subsequent "scoping" of that package, but that's just one(1) use case; this functionality is also useful for packages which do have releases via git tags, but have not been registered with npm, for example!

- new generated data is appended to the top of an existing file.
- the tool searches the local directory for any file matching the 'changelog*' glob search crriterion and picks the first one it finds. When no file is found, 'CHANGELOG.md' is assumed as a default destination. This behaviour ensures that various incantations of the name do not cause any trouble in `-a`-updating, while any newly created CHANGELOG will have a github-friendly name: "CHANGELOG.md"

…of proper OAuth transfer to github

…lint, test, etc.

… (I don't trust the 'single releae changelog entries only'-behaviour yet...)

Some non-open-source npm packages like
https://www.npmjs.com/package/bitmovin-player contain repository fields
with empty URLs. This commit handles this case which other wise would
throw an error.

# Conflicts:
#	package.json

…S and active NodeJS versions in package.json (>= 8.0)

…me promise as any positive result.

- fix: typos in test descriptions
- add tests for the fix to prevent regression

…me promise as any positive result.

- fix: typos in test descriptions
- add tests for the fix to prevent regression

# Conflicts:
#	test/changelog.test.js

# Conflicts:
#	package-lock.json
#	package.json
#	test/changelog.test.js