Struct Config 

pub struct Config {
    pub module_id: String,
    pub logs_dir: PathBuf,
    pub bound: Vec<(String, String)>,
    pub spec: Option<String>,
    pub retention: SegmentRetention,
    pub redactor: Option<Arc<Redactor>>,
    pub clock: Option<Arc<dyn Fn() -> SystemTime + Send + Sync>>,
}

Logger configuration for one process.

Fields

module_id: String

Stable fleet module identifier: the root of every logger name and the first part of every segment’s file name.

logs_dir: PathBuf

The directory the segments live in. Fleet callers never assemble this: Config::for_module derives it from the module id through the same resolver every store uses, so a log cannot land beside the wrong data. Public for tests and for the daemon’s own run/logs/ lane, which is not a module data directory.

bound: Vec<(String, String)>

Fields bound for the whole process and rendered in the bracket on every line, in this order. A harness-hosted plugin binds harness=<name> here: after r2 every lane shares one segment and this is what tells them apart.

spec: Option<String>

CK_LOG override; None reads the process environment.

retention: SegmentRetention

Age window and oversize alarm for the segments.

redactor: Option<Arc<Redactor>>

Optional module redactor, applied after fleet credential redaction.

clock: Option<Arc<dyn Fn() -> SystemTime + Send + Sync>>

Optional clock override for deterministic callers and tests.

Implementations

impl Config

pub fn in_dir(module_id: &str, logs_dir: impl Into<PathBuf>) -> Self

A configuration whose segment directory the caller already resolved: CK_LOG from the environment, fleet default retention, nothing bound. This is the path for a consumer that owns its module data directory through its own pinned store crate and must not pull a second copy of it through this one (see the store-paths feature).

pub fn for_module(module_id: &str) -> Self

The fleet configuration for a supervised module’s own process: segments under <module data dir>/logs/, CK_LOG from the environment (the daemon injects it at spawn), fleet default retention, nothing bound.

pub fn for_plugin(module_id: &str, harness: &str) -> Self

Config::for_module with harness=<harness> bound on every line, for a plugin running inside a harness process.

pub fn from_env() -> Result<Self, InitError>

The configuration a supervised module can build with no arguments, because the daemon already injects everything it needs at spawn: SUBC_MODULE_ID (for launch attestation), CK_LOG, and the retention knobs as CK_LOG_MAX_AGE_DAYS / CK_LOG_ALARM_SEGMENT_MB. This is the zero-argument path the r2 spec requires so that reaching for the crate costs no more than reaching for eprintln!.