Keyed logger used to define processing pipelines on a log stream.

You can download the code via the usual go utilities:

go get github.com/datacratic/goklog

To build the code and run the test suite along with several static analysis tools, use the provided Makefile:

make test

Note that the usual go utilities will work just fine but we require that all commits pass the full suite of tests and static analysis tools.

While klog does offer nifty features, it was generally designed to compliment existing log framework. It does so by defining a very generic set of abstractions (lines, printers and chained printers) which can be used to manipulate the log output prior to handing off to an existing log framework.

Examples are available in the following test suites:

• Pipeline: Basic library usage along with how to setup a pipeline.
• REST pipeline: How to setup a REST enabled pipeline.

In this section we'll give a quick overview of the various klog printers.

Filter can filter out/in lines based on a full-text, prefix or suffix matching rules. Note that this implementation is generally more expensive then filtering strategies used other logging libraries but its performance are generally acceptable and the flexbility is always useful.

Filters are also available through a REST interface which allows for the live manipulation of logs in real-time.

Dedup is used to aggregate the consecutive identical lines for a given key into a single line. This is useful to avoid flooding the logs with endless identical messages.

Ring logs all the received lines into a fixed size ring buffer in a lock-free manner. It's mostly useful when coupled with its REST interface which enables the logs of a given service to be queried remotely.

The source code is available under the Apache License. See the LICENSE file for more details.