[nickpresta]

I find well written changelogs far more useful than relying on encoding meaning in a version string.

I really like http://keepachangelog.com as a guideline for writing changelogs.

[RickHull]

This doesn't help with dependency management. Changelogs, no matter how well written, cannot replace semver in e.g. Ruby / Rubygems / Bundler version constraints:

https://robots.thoughtbot.com/a-healthy-bundle

[kelnage]

Is there any harm in doing both? I agree that good changelogs are incredibly helpful - but I can also see how consistent version numbering can be useful too. If nothing else, it can allow easier automated dependency management vs. parsing the contents on a changelogs.

[nickpresta]

I'm of the opinion that if you're relying on semver for automated depenency management (like npm), you're doing it horribly wrong.

Pin your dependencies to specific versions, read changelogs and make informed decisions.

Don't rely on someone bumping patch versions correctly to keep your software stable.

I tend to agree with Jeremy Ashkenas on this issue: https://gist.github.com/jashkenas/cbd2b088e20279ae2c8e