Struct EditionDefaults 

pub struct EditionDefaults {
    pub default_max_episode_steps: i64,
    pub trial_index_option_key: &'static str,
    pub driver_owned_reset_modes: &'static [AutoresetMode],
    pub success_info_keys: &'static [&'static str],
    pub conformance_warning_info_key: &'static str,
}

The edition-governed defaults one Edition promises: the values a participant would otherwise hardcode, lifted here so a future default change is a new row rather than an edit to a code path a sealed edition already depends on. One row per arm, returned by defaults; a sealed row is never edited.

Fields

default_max_episode_steps: i64

Per-episode step bound applied when the session sets no explicit max_episode_steps and the runtime owns resets, so a broken termination condition surfaces as a truncation instead of hanging the run.

Not named in the prose of docs/editions/2026.06.md, but observable in the episode ledger the edition’s “Episode Accounting” section governs (“When a sub-environment reports terminated or truncated, its episode completes”), and pinned by the 05-default-truncation-bound fingerprint.

Mirrored by _MAX_STEPS_PER_EPISODE in python/rlmesh/src/rlmesh/_models/_eval.py, which bounds the Python Session loop identically. The two loops must agree, so an edition that changes this row changes that constant in the same commit.

trial_index_option_key: &'static str

The single reserved ResetRequest.options key this edition defines.

docs/editions/2026.06.md, Reset: “This edition reserves exactly one, trial_index: the 0-based ordinal of the episode being started, an integer for a single lane or a per-lane list aligned to the lanes being reset.”

driver_owned_reset_modes: &'static [AutoresetMode]

The autoreset modes under which the RUNTIME owns lane restarts (it issues the Resets), as opposed to the env rolling its own lanes.

docs/editions/2026.06.md, Episode Accounting: “AUTORESET_MODE_DISABLED: only an explicit Reset restarts a lane.” and “AUTORESET_MODE_UNSPECIFIED: treated as DISABLED.”

success_info_keys: &'static [&'static str]

The final_info keys this edition reads an episode’s task outcome from, in priority order: the first key present decides, and a numeric value coerces by truthiness (1/1.0 → true).

Which keys count is an edition-governed promise — an env that reports its outcome under one of these names has it surface on the episode summary and the EpisodeCompleted event; one that uses any other name does not. Gymnasium’s is_success leads; success and task_success follow as the two spellings the ecosystem also ships.

The Python Session loop reads the outcome from the same info with bool(info[key]) (_success_from_info in python/rlmesh/src/rlmesh/_models/_eval.py); an edition that changes this row revisits that sweep in the same commit.

conformance_warning_info_key: &'static str

The reserved info-map key a served env reports value-conformance warnings under.

docs/editions/2026.06.md, Value Conformance: “A conformance warning is reported in-band under the reserved rlmesh.conformance.warning key in the info map returned by reset and step, at most once per (deviation kind, value path) per session.”

Trait Implementations

impl Clone for EditionDefaults

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for EditionDefaults

impl Debug for EditionDefaults

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for EditionDefaults

impl PartialEq for EditionDefaults

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for EditionDefaults