Generated pre-release part of SemVer differs from SemVer spec

[Xiangze-Li]

Description

When using SemVer schema and bumping to pre-release version, the version generated is like 1.0.0-a0.

This differs from examples from SemVer spec 9, which is 1.0.0-alpha.0, using "alpha" instead of "a", and having a dot between "a" and the sequential number.

Current behavior may break the version comparison when the number has more than 1 digits.
For example, 1.0.0-a10 is less than 1.0.0-a2 according to spec 11, because a10 is less than a2 lexically.

Steps to reproduce

1. Setup commitizen using SemVer schema
2. Run cz bump --pr alpha

Current behavior

Bumped into x.y.z-a0

Desired behavior

Bumped into x.y.z-a.0 or x.y.z-alpha.0

Screenshots

No response

Environment

Commitizen Version: 3.18.3
Python Version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0]
Operating System: Linux

[woile]

Did you configure the version scheme?
https://commitizen-tools.github.io/commitizen/bump/#version_scheme

[Xiangze-Li]

Did you configure the version scheme? https://commitizen-tools.github.io/commitizen/bump/#version_scheme

I set this field (when initializing) in .cz.toml as version_scheme = "semver".

Actual behavior is exactly same with what stated in the linked manual, which disagrees with SemVer spec.

[Xiangze-Li]

I just realized this behavior is OK under SemVer 1.0.0.

Maybe it's nice to seperate the two SemVer more clearly?

[woile]

What do you mean by "the two semver"?

[Xiangze-Li]

What do you mean the two semver?

SemVer has two major versions now. Version 2 has changed the spec of the pre-release part.

[woile]

mm we may need a new version scheme: semver_v2
Thoughts on this @Lee-W @noirbizarre ?

[krumlo]

Second this, I'd really enjoy semver 2.0 support!

[noirbizarre]

Hum, makes sense.
I need to review quickly differences with v1, but it should be easy with the versioon providers

[noirbizarre]

There it is, implemented in #1072.

Note that SemVer2 is underspecified on the prerelease part (I include development releases as the concept does not exist in SemVer2) and some choices has been made, staying close to the examples (only example are using alpha, beta), the specification itself is just about the format and the precedence rules, it does not specify the standard parts (alpha, beta, dev... which might also be a, benddev`).

(This point is one of the reason than Python community dropped SemVer for PEP440)

[Xiangze-Li]

Many thanks!

[krumlo]

@noirbizarre very quick feedback, thanks a bunch!!