2.8 KiB
2.8 KiB
- document streams better in the readme
- buffered writes to increase speed: Yunong said he'd work on a patch:
- perhaps this would be a "buffered: true" option on the stream object
- then wrap the "stream" with a local class that handles the buffering
- "canWrite" handling for full streams. Need to buffer a la log4js
- test file log with logadm rotation: does it handle that?
- test suite:
- test for a cloned logger double-
stream.end()
causing problems. Perhaps the "closeOnExit" for existing streams should be false for clones. - test that a
log.clone(...)
adding a new field matching a serializer works and that an existing field in the parent is not re-serialized.
- test for a cloned logger double-
- a "rolling-file" stream: but specifically by time, e.g. hourly. (MarkC requested)
- ringBuffer stream
someday/maybe
log.close
to close streams and shutdown andthis.closed
process.on('exit', log.close)- bunyan cli: -c COND args a la json
- bunyan cli: more layouts (http://logging.apache.org/log4j/1.2/apidocs/org/apache/log4j/EnhancedPatternLayout.html) Custom log formats (in config file? in '-f' arg) using printf or hogan.js or whatever. Dap wants field width control for lining up. Hogan.js is probably overkill for this.
- syslog: Josh uses https://github.com/chrisdew/node-syslog streams: [ ... { level: "warn", type: "syslog", syslog_facility: "LOG_LOCAL1", // one of the syslog facility defines syslog_pid: true, // syslog logopt "LOG_PID" syslog_cons: false // syslog logopt "LOG_CONS" }
- bunyan "compact" or "light", '-l'? Something like. Or pehaps this (with color) could be the default, with '-l' for long output. 13:51.340 [src.js:20#Wuzzle.woos] WARN: This wuzzle is woosey.
- get Mark to show me dtrace provider stuff and consider adding for logging, if helpful.
- add option to "streams" to take the raw object, not serialized. It would be a good hook for people with custom needs that Bunyan doesn't care about (e.g. log.ly or hook.io or whatever).
- split out
bunyan
cli to a "bunyan" or "bunyan-reader" or "node-bunyan-reader" as the basis for tools to consume bunyan logs. It can grow indep of node-bunyan for generating the logs. It would take a Bunyan log record object and be expected to emit it. - serializer
request_id
that pulls it from req?log.info({request_id: req}, "hi")
- serializer support:
- restify-server.js example -> restifyReq ? or have
req
detect that. That is nicer for the "use all standard ones". Does restify req have anything special? - differential HTTP client req/res with server req/res.
- restify-server.js example -> restifyReq ? or have
- statsd stream? http://codeascraft.etsy.com/2011/02/15/measure-anything-measure-everything/ Think about it.