yup, that's why the scheme i'm describing is becoming the most common standard, it's practical, it makes the activity visible, and the more people adopt it for their projects the less thinking you have to do when you ask "ok, so how do i build this from source" - it should be boilerplate on readme.md's to have basic instructions about tagging and such...

i stick with semver so that's another thing also, with semver you pretty much know that vX.X.0 should be stable, eg v1.2.0 should be a stable minor release and v2.0.0 should be a stable major, and anything with numbers after it is either a fix or some/all of a feature added, and if the commit has no tag then you can expect it to break