Don't touch package versions if package seems not to follow semver

[do3cc]

While this package nowhere claims it supports semver, I just declare that it does so. Some package maintainers believe that as long as there is no patch version, there is no need for a patch version. So there may be packages like 1.0 and 1.2.9. In case of 1.0, increasing the patch number with -2 will result in getting this change: 1.0 -> 1.2.9 This brings changes not really wanted.

[mgedmin]

I don't understand your meaning.

[do3cc]

From the docs: 2 gets the highest minor version (x.y.Z). Using level 2, you can automatically retrieve all bugfix versions of a buildout

What I do: checkversions ... -l 2... What I expect: version updated from 1.0 to 1.0.z (With z the highest existing number) or unchanged. What I got: version updated from 1.0 to 1.y.z (with y and z the highest existing number, 1.2.9 in my case)

[do3cc]

My expectations come from following the ideas from semver.org. z3c.checkversions is from a time before semver.org became widely used, but I think z3c.checkversions follows roughly the same ideas of semver.org

[mgedmin]

Thank you, now I understand.

Yeah, feels wrong to me too.

[do3cc]

Would it be ok to you if I reword the documentation to use terms from semver and also reword some of the arguments? -1 -> -minor -2 -> -patch... In a PR, of course

[mgedmin]

You mean create --major/--minor/--patch (or --bugfix) as aliases for --level=1/2/3? :+1:

-1 means something unrelated, AFAIU.