Performance & caching

Atemporal caches parsing, formatting, and repeated difference calculations internally. Consumer code should use the public factory diagnostics rather than importing internal cache classes. The API returns snapshots, so modifying a returned object never changes library state.

import atemporal from "atemporal";

atemporal.prewarm({
  formatPatterns: ["YYYY-MM-DD", "YYYY-MM-DD HH:mm"],
});

const diagnostics = atemporal.getDiagnostics();
console.log(diagnostics.temporal);
console.log(diagnostics.caches.parsing);
console.log(diagnostics.caches.formatting);
console.log(diagnostics.caches.diff);

Public operations

Method	Use
atemporal.getDiagnostics()	Get a detached, serializable snapshot of Temporal runtime and cache metrics.
atemporal.clearCaches()	Drop cached parsing, formatting, and diff entries without changing defaults or counters.
atemporal.resetDiagnostics()	Clear caches and reset parser/formatter counters without changing the configured locale or time zone.
atemporal.prewarm({ formatPatterns })	Initialize formatting and compile the supplied application patterns during startup.

Use prewarm only for a small, stable set of hot patterns. It is usually best called once during application initialization, not for every request or render.

Boundaries and measurements

Diagnostics are intentionally observational: no cache or parsing engine is public API. Do not mutate the snapshot, retain assumptions about its internal counters, or import atemporal/src/... in application code.

For release-size and benchmark policy, see the repository's generated reports. Application bundles depend on the bundler, your imports, and whether the Temporal polyfill is included; treat those as separate measurements from the distributed core size.