Including machine-parseable formats for search and indexing. The Logger.WithFields and Logger.WithValues methods, and provides flexible formatting options for rendering these values, Cue simplifies collection of these details via It's often valuable to know under what conditions application events trigger: the account thatĪccessed a page, flags associated with the account, country of origin, etc. Where that error is sent should beĪ configuration detail and shouldn't require additional service-specific API calls. I feel it's important to be able to log/collect application errors without needing to explicitly specify where to send them.Ĭalling log.Error/log.Recover should be enough to trigger the error report. However, there's nothing particularly special about how error reporting services work. They provide immediate notification and visibility into bugs that your end users are encountering. Error reporting services are incredibly valuable for productionĪpplications. Built-in support for error reporting services.I didn't start with the intention of creating a new library, but I struggled to find an existing logging library that There are quite a few Go logging libraries that already exist. Panic( 'No program', 'Whoops, there's no program to run!') '/bobziuchkovski/cue' '/bobziuchkovski/cue/collector' 'os' 'syscall'Ĭue. See cue/format.JSONMessage and cue/format.JSONContext. The context data may also be formatted as JSON for machine parsing The formatting may be changed by passing a different formatter to either collector.įor details. This is useful for log rotation.Īdditional context is added via the. Receives SIGHUP, the file will be reopened. This example logs to both the terminal (stdout) and to file. Gracefully handles and recovers from failures.On a 2015 MacBook pro, these calls take ~16ns/call, meaning tens of millions of no-op calls may be serviced DEBUG in production) are quickly pruned via atomic operations.

Match a registered Collector threshold (e.g. Cue uses atomic operations to avoid lock contention and reusesįormatting buffers to avoid excessive memory allocation and garbage collection. If the end user doesn't configure log collection, logging calls are silently dropped. Log collection is explicitly opt-in, meaning cue is safe to use within

The rest of the API is centered around log output configuration.įully asynchronous (guaranteed non-blocking) log collection. You only need to know cue.NewLoggerĪnd the cue.Logger interface to get started. Release, the API is guaranteed to remain backwards compatible.Ĭue makes use of sync/atomic.Value and thus requires Go 1.4.x or later. Minor breaking changes may occur prior to the 1.0 release. Services such as Honeybadger, Loggly, Opbeat, Rollbar, and Sentry. Thorough test coverage and supports logging to stdout/stderr, file, syslog,Īnd network sockets, as well as hosted third-party logging and error/reporting Cue implements contextual logging with 'batteries included'.

Godocs sentry