what should our behavior be with lerna

[bcoe]

I've been investigating moving a few of my repos over to lerna. Automated CHANGELOG generation has become an important part of my workflow -- helping me quickly bisect bad releases.

I'd like to raise the question what should our behavior be with lerna?

• should we be creating CHANGELOG for each package in the lerna repo?
• should we generate a top level CHANGELOG for the lerna repo instead? (this seems like it might fit better with our flow).

I've created a testing repo which I'm going to start noodling in:

https://github.com/bcoe/standard-version-lerna-test

@hzoo, @Tapppi do you have any thoughts on this subject -- once we figure out a happy path for automatic CHANGELOG mangement in lerna, maybe this is even something we could add to the lerna project?

[hzoo]

real quick: we setup https://github.com/lerna/lerna-changelog which is a simple script that uses github tags to label prs but from the original pr there was talk of using coventional changelog, and a fork of semantic release etc for lerna lerna/lerna#29 (comment)

[bcoe]

@hzoo I like this 👍 I had a chance to experiment a bit more; I think I'm going to extend on our tag parser slightly, so that it supports tags in the format: foo-module@1.3.0, which will allow me to create a changelog on a package by package basis.

[bcoe]

see: lerna/lerna#665