Programatically determining if a project is using semantic versioning. about semver.org HOT 7 CLOSED

How about using a special character like §X.Y.Z on the first version/subsequent versions that uses semantic notation? I see problems in just requiring vX.Y.Z to be in the history because some people might be doing that already without regard to its meaning in this context.

from semver.org.

Very good. The clarification that the semver tag must point to the same changeset as a vX.Y.Z tag is missing from the website.

from semver.org.

The biggest problem here is that I don't want to require the use of a versioning system in order to be SemVer compliant.

from semver.org.

Just b/c someone says they are using semver doesn't mean they have done so, and especially to perfection. I think we need to be intellectually honest here and see semver for what it is, a strong set of guidelines. It is not something that can be readily tested or verified about a project. So I doubt any tools can really make much use of the fact.

That being said, the proper approach would be a field in the project metadata (e.g. gemspec) like version_policy set to semver.

from semver.org.

From the package manager side of things, I only want to see three-part version numbers (with an optional build suffix) all separated by a ..

From an automated-tool/SCM point of view, only recognizing tags which begin with a v is the easiest route. This allows tools to ignore any other non-semver tags (0.2) or date tags (4.20.2010).

from semver.org.

The bottom line is that if it's too hard to be compliant, then nobody well even bother. The spec is lightweight on purpose, to make sure the philosophy is conveyed. For most projects, absolute compliance is impossible because they will never properly specify a Public API. That is fine. As long as the philosophy of SemVer is being understood, then the spec is accomplishing my goal.

from semver.org.

SemVerTag will be removed from the spec, so this is no longer an issue. I don't think automatic detection of SemVer compliant packages was ever viable in the first place.

from semver.org.