Logging API

This page is the API reference for EdgeCommons logging: the logger type in each SDK, how you obtain one, the log methods and level predicates, the structured UNS log publisher, the configuration type, the configurator entry points, and the per-language format tokens. For the concepts behind these — how levels are honored, the stdout-JSON sink, file rotation, log-bus publishing, and hot reload — see the Logging guide.

There is no gg.logging() accessor in any SDK. The framework configures your language’s logging framework when you build your EdgeCommons instance; component code obtains a logger from the native facility and writes to it. The optional structured bus publisher is exposed separately as getLogs() in Java and logs() in Python, Rust, and TypeScript.

At a glance

SDK	Obtain a logger	Logger type	Log methods	Predicates	Format key
Java	`LoggerFactory.getLogger(...)`	`com.mbreissi.edgecommons.logging.Logger` (interface)	`trace` `debug` `info` `warn` `error`	`isTraceEnabled()` … `isErrorEnabled()`	`java_format`
Python	stdlib `logging.getLogger(name)`	stdlib `logging.Logger`	`debug` `info` `warning` `error` `critical`	`isEnabledFor(level)`	`python_format`
Rust	`tracing` macros (no factory)	n/a (macros)	`trace!` `debug!` `info!` `warn!` `error!`	`tracing` level checks	`rust_format`
TypeScript	`logger` / `getLogger(name)`	`Logger` (class)	`debug` `info` `warn` `error`	none exported	`ts_format`

Obtaining a logger

import com.mbreissi.edgecommons.logging.Logger;
import com.mbreissi.edgecommons.logging.LoggerFactory;

Logger byClass = LoggerFactory.getLogger(MyApp.class);
Logger byName  = LoggerFactory.getLogger("com.example.MyApp");

LoggerFactory is a static factory:

public final class LoggerFactory {
    public static Logger getLogger(Class<?> clazz);
    public static Logger getLogger(String name);
}

The factory auto-detects the underlying framework once at class-load time, preferring SLF4J, then Log4j2, then JUL. You can also bypass the facade and use Log4j2 directly (LogManager.getLogger(...)) — the library has already reconfigured the Log4j2 root logger.

import logging

log = logging.getLogger(__name__)

There is no EdgeCommons logger factory in Python. The library configures the root logger, so the stdlib logging.getLogger(name) is all you need. Per-logger levels from logging.loggers apply by the name you pass here.

use tracing::{debug, info, warn, error};

There is no getLogger in Rust. Components emit through the tracing macros; the library installs the global tracing-subscriber once during build(). The macro’s target defaults to the current module path, which is what per-logger loggers directives match against.

import { logger, getLogger } from "@edgecommons/edgecommons";

logger.info("using the process-wide root logger");   // name "edgecommons"

const dbLog = getLogger("app.db");                    // named, cached, hierarchically leveled

logger is the process-wide root Logger (named "edgecommons"). getLogger(name) returns a named, cached Logger whose level is resolved by longest dotted-prefix match against logging.loggers.

The logger type

com.mbreissi.edgecommons.logging.Logger is an interface. Each of trace/debug/info/warn/error has three overloads, plus a level predicate per level:

public interface Logger {
    void trace(String message);
    void trace(String message, Object... args);      // SLF4J-style {} placeholders
    void trace(String message, Throwable throwable);
    // ... identical trio for debug, info, warn, error ...

    boolean isTraceEnabled();
    boolean isDebugEnabled();
    boolean isInfoEnabled();
    boolean isWarnEnabled();
    boolean isErrorEnabled();
}

The (String, Object...) overload uses SLF4J-style {} placeholders; the (String, Throwable) overload attaches an exception. There is no fatal() method on this interface.

Python uses the standard-library logging.Logger returned by logging.getLogger(name) — EdgeCommons does not define its own logger type. The standard methods are:

log.debug("payload bytes: %d", len(payload))
log.info("publish interval set to %sms", interval)
log.warning("retrying after error", exc_info=True)
log.error("query failed")
log.critical("fatal condition")                   # FATAL maps to CRITICAL

if log.isEnabledFor(logging.DEBUG):               # the stdlib level predicate
    log.debug(expensive())

These are stdlib semantics; EdgeCommons only configures the root logger, level, format, and sinks.

Rust has no logger object — components use the tracing macros, and EdgeCommons installs the subscriber that decides what is emitted:

tracing::trace!(detail = ?value, "trace line");
tracing::debug!(payload_bytes = payload.len(), "received payload");
tracing::info!(publish_interval = interval, "configuration changed");
tracing::warn!(error = %err, "retrying after error");
tracing::error!("query failed");

Structured key/value fields are preserved verbatim in the JSON sink. Level “predicates” are handled by tracing itself (the subscriber and EnvFilter); there is no EdgeCommons-provided isEnabled helper.

Logger is an exported class. Construction is normally via getLogger(name) / the logger singleton rather than new:

export class Logger {
  readonly loggerName: string;
  constructor(name?: string);                       // defaults to "edgecommons"

  debug(message: string, error?: unknown): void;
  info(message: string, error?: unknown): void;
  warn(message: string, error?: unknown): void;
  error(message: string, error?: unknown): void;

  refresh(): void;                                  // recompute level/format/sink from config
  setLevel(level: Level): void;                     // Level enum is internal — see note
  setFormat(format: string | undefined): void;      // ts_format template; undefined → default
  setFileWriter(writer: RotatingFileWriter | undefined): void;
}

There are only four log methods (debug/info/warn/error) — no trace/fatal, and no level-predicate methods. Each method takes an optional second error?: unknown. In text mode warn/error write to stderr and the rest to stdout; in JSON mode all levels go to stdout.

The logging configuration type

These types model the logging config section and drive the configurator. They are framework-facing — you rarely construct them yourself — but they expose the resolved values via getters.

com.mbreissi.edgecommons.config.LoggingConfiguration parses the logging JSON object and exposes typed getters:

LoggingConfiguration(JsonObject json);

org.apache.logging.log4j.Level getLevel();          // default INFO
String  getFormat();                                 // java_format or DEFAULT_FORMAT
boolean isFormatExplicitlySet();                     // true if java_format was provided
boolean isFileLoggingEnabled();
String  getLogFilePath();                            // == fileLogging.filePath
String  getMaxFileSize();                            // default "10MB"
int     getBackupCount();                            // default 5
java.util.Map<String, Level> getLoggerLevels();      // from logging.loggers
boolean isGlobalControlEnabled();                    // default false
LogPublishConfiguration getPublishConfig();          // logging.publish

Defaults: level INFO, format %d{yyyy-MM-dd HH:mm:ss} [%-5p] %-25.25c{1}(%4L): %m%n, maxFileSize "10MB", backupCount 5. Obtain the live instance via ConfigManager.getLoggingConfig().

edgecommons.config.enhanced_logging_config.EnhancedLoggingConfiguration parses the section, applies it, and exposes getters:

EnhancedLoggingConfiguration(
    logging_config=None,            # the logging dict
    platform_default_format=None,   # e.g. "json" on KUBERNETES
    correlation=None,               # thing/pod/namespace/node fields
)

cfg.configure_logging(config_manager=None)   # clears root handlers, installs sinks
cfg.get_level(); cfg.get_format(); cfg.is_json_sink()
cfg.is_file_logging_enabled(); cfg.get_log_file_path()
cfg.get_logger_levels(); cfg.is_global_control_enabled()
cfg.get_publish_config(); cfg.to_dict()

Defaults: level logging.INFO, format %(asctime)s [%(levelname)s] %(name)s: %(message)s. The companion JsonLogFormatter(logging.Formatter) renders the JSON sink (UTC timestamps via time.gmtime). Obtain the live instance via ConfigManager.get_logging_config().

Rust has no separate configuration class — logging::init/logging::reconfigure read the parsed logging section off config.parsed.logging:

// fields read from config.parsed.logging
level: String,                          // default "INFO"
rust_format: Option<String>,            // token template or "json"
loggers: HashMap<String, String>,       // target -> level (EnvFilter directives)
file_logging.enabled: bool,
file_logging.file_path: String,
file_logging.max_file_size(): String,   // default "10MB"
file_logging.backup_count(): u32,       // default 5
publish: LoggingPublishConfig,          // logging.publish

There are no getters to call; the values are consumed directly by the configurator.

TypeScript has no exported configuration class — initLogging/reconfigureLogging read the parsed logging section off config.parsed.logging:

// fields read from config.parsed.logging
level: string;                          // default "INFO"
tsFormat?: string;                      // token template or "json"
loggers: Record<string, string>;        // logger name -> level
fileLogging.enabled: boolean;
fileLogging.filePath: string;
fileLogging.maxFileSize(): string;      // default "10MB"
fileLogging.backupCount(): number;      // default 5
publish: LoggingPublishConfig;

The platform default (e.g. json on Kubernetes) is supplied through LoggingOptions:

interface LoggingOptions {
  formatDefault?: string;   // platform-profile default format
  env?: Env;                // environment lookup for correlation fields
}

Configurator and reconfiguration entry points

The framework calls these at startup and on hot reload. They are documented here for completeness; component authors do not normally invoke them.

// on ConfigManager — the live runtime configurator
public void reconfigureLogging();             // rebuilds the Log4j2 config from LoggingConfiguration
public LoggingConfiguration getLoggingConfig();

reconfigureLogging() runs at init and again on hot reload via the registered LoggingConfigChangeListener. It always installs a SYSTEM_OUT console appender, optionally a size-rotated RollingFile appender, sets the root level, and adds a per-logger LoggerConfig (with additivity=false) for each logging.loggers entry. When globalControl=true it delegates to GlobalLoggingManager, which rebuilds the entire Log4j2 configuration.

EnhancedLoggingConfiguration(...).configure_logging(config_manager=None)

configure_logging() clears the root handlers, sets the root level, installs a StreamHandler (JSON or text formatter), and — only when fileLogging.enabled and not under the JSON sink — a logging.handlers.RotatingFileHandler(maxBytes, backupCount). The ConfigManager rebuilds and calls this on hot reload. Note: Python parses globalControl (is_global_control_enabled()) but configure_logging() never acts on it.

use edgecommons::{config::model::Config, logging};

logging::init(&config, profile_format_default);   // install the global subscriber once (idempotent)
logging::reconfigure(&config);                     // swap the level EnvFilter live

init takes the parsed Config and an Option<&str> platform default (pass Some("json") for the Kubernetes profile). It installs the global tracing-subscriber exactly once. reconfigure swaps only the level filter live — the format and file layers are fixed at init, so a rust_format or fileLogging change requires a restart. LoggingReconfigurer implements ConfigurationChangeListener to call reconfigure on reload.

import { initLogging, reconfigureLogging, LoggingReconfigurer } from "@edgecommons/edgecommons";

initLogging(config, options?);        // (config: Config, options?: LoggingOptions)
reconfigureLogging(config);           // re-run the full configurator

initLogging configures all Logger instances from config.parsed.logging (level, per-logger levels, format/JSON mode, and the rotating file writer). reconfigureLogging re-runs the full configurator on hot reload and can swap the file writer live. LoggingReconfigurer implements ConfigurationChangeListener and forwards reload events to reconfigureLogging.

Structured log bus publisher

The log UNS class is library-owned and reserved. When logging.publish.enabled is true, EdgeCommons publishes structured records to:

ecv1/{device}/{component}/log/{level}

Each message uses envelope header { name: "log", version: "1.0" } and body schema edgecommons.log.v1. The core body fields are timestamp, level, logger, message, and sequence; optional fields include thread, fields, error, truncated, and dropped. Application code still cannot publish raw messages to .../log/... through the public messaging API; the log publisher uses the internal reserved publish seam.

LogService logs = gg.getLogs();
logs.publish(LogRecord.builder()
    .withLevel("INFO")
    .withLogger("app")
    .withMessage("started")
    .addField("instance", "main")
    .build());
logs.flush(Duration.ofSeconds(2));
LogStats stats = logs.stats();

from edgecommons.logs import LogRecord

gg.logs().publish(LogRecord(
    level="INFO",
    logger="app",
    message="started",
    fields={"instance": "main"},
))
gg.logs().flush(timeout=2.0)
stats = gg.logs().stats()

use edgecommons::logs::{LogLevel, LogRecord};

gg.logs().try_publish(
    LogRecord::builder(LogLevel::Info, "app", "started")
        .field("instance", serde_json::json!("main"))
        .build(),
)?;
let stats = gg.logs().stats();

await gg.logs().publish({
  level: "INFO",
  logger: "app",
  message: "started",
  fields: { instance: "main" },
});
await gg.logs().flush();
const stats = gg.logs().stats();

Automatic capture is controlled by logging.publish.captureNative and logging.publish.captureConsole. Native capture observes the logging framework that the SDK wires: Log4j2/JUL/SLF4J in Java, stdlib logging in Python, tracing in Rust, and EdgeCommons Logger in TypeScript. Console/stdout capture is runtime-dependent; the config field is shared, but not every SDK can intercept arbitrary third-party output.

Format tokens

Each SDK reads only its own format key, written in its own syntax; a token from one language is meaningless to the others. Setting the value to the literal json (case-insensitive) selects the stdout-JSON sink instead of a text layout.

SDK	Config key	Syntax	Default (when unset)
Java	`java_format`	Log4j2 `PatternLayout`	`%d{yyyy-MM-dd HH:mm:ss} [%-5p] %-25.25c{1}(%4L): %m%n`
Python	`python_format`	`logging.Formatter` string	`%(asctime)s [%(levelname)s] %(name)s: %(message)s`
Rust	`rust_format`	token template	none — the `tracing-subscriber` `fmt` layer (the `{timestamp} {level} {component}/{componentName} {target} {message}` tokens apply only when `rust_format` is set explicitly)
TypeScript	`ts_format`	token template	`{timestamp} [{level}] {message}`

Token-template placeholders differ between Rust and TypeScript: Rust supports {timestamp}, {level}, {target}, {message} (it uses {target}), while TypeScript supports {timestamp}, {level}, {logger}, {message} (it uses {logger}). The effective format is resolved with the precedence: explicit *_format token, then the platform-profile default (json on Kubernetes), then the SDK’s built-in text default.

Example: get a logger and log

import com.mbreissi.edgecommons.logging.Logger;
import com.mbreissi.edgecommons.logging.LoggerFactory;

Logger log = LoggerFactory.getLogger(MyApp.class);   // or getLogger("my.logger")

log.info("publish interval set to {}ms", interval);  // SLF4J-style {} args
log.warn("retrying after error", throwable);         // (String, Throwable) overload
if (log.isDebugEnabled()) {
    log.debug("payload bytes: {}", payload.length);
}

import logging

log = logging.getLogger(__name__)

log.info("publish interval set to %sms", interval)
log.warning("retrying after error", exc_info=True)
log.debug("payload bytes: %d", len(payload))

use tracing::{debug, info, warn};

info!(publish_interval = interval, "configuration changed");
warn!(error = %err, "retrying after error");
debug!(payload_bytes = payload.len(), "received payload");

import { logger, getLogger } from "@edgecommons/edgecommons";

logger.info("publish interval set to " + interval + "ms");

const dbLog = getLogger("app.db");
dbLog.debug("db detail");                  // honors logging.loggers["app.db"]
dbLog.error("query failed", err);          // (message, error?) signature