A schema is the recipe that turns version variables into a version string. Every zerv
output, from both version and flow, is rendered through one.
Architecture
All schemas resolve to the internal ZervSchema struct with three required components:
core - primary version components (e.g., [Major, Minor, Patch] for SemVer)extra_core - additional version components (e.g., pre-release, post-release, dev)build - build metadata components (e.g., commit hash, branch name, build info)
core and extra_core form the sortable identity of the version; build carries
traceability metadata after a +.
Schema resolution: preset schemas (standard-base, calver-*, etc.) are predefined
ZervSchema objects that adapt based on repository state. RON schemas are parsed from text
into the same ZervSchema structure, providing identical functionality with custom
definitions.
Presets
22 presets ship built-in, 11 per family:
standard-* - SemVer-shaped cores. Suffixes add parts:
-context (build context), -prerelease, -post, -dev, and combinations like
standard-base-prerelease-post-dev-contextcalver-* - the same suffix system over a date-based core
(YYYY.MM.DD-<patch>), e.g. calver-base-prerelease-post-dev-context
Custom RON schemas
Write a schema inline with --schema-ron. Each component is a list of variables: var(...)
references version fields, str("...") injects literals:
The last example demonstrates date formatting with var(ts("YYYY")): any chrono format
string works inside ts(...).
Common uses:
v-major tags - (core:[var(Major)], extra_core:[], build:[]) renders bare 1, 2;
add --output-prefix v --output-format pep440 for v1, v2- Minimal SemVer - drop
build for clean, sortable versions in release artifacts
- CalVer releases - date-based cores for products versioned by ship date
Schema selection
flow takes the same --schema / --schema-ron flags with its own preset output shapes;
see the flow CLI.
Where to go next
Last modified on August 25, 2026
