0.9.11 More configuration options, optimisations and documentation

#60 - Added ability to configure size of thread local buffer to avoid message fragmentation (see here and here).

#53 - Optimised generation of labels that contain MDC/ctx reference that is dynamically resolved on every message. Replaced usage of log4j2 interpolation logic with allocation free one.

(this one was not release due to problems with maven oss repository), try with 0.9.12.

0.9.10 Per message substitution of ${ctx:} lookups

All environment, system etc. variable references are be replaced by their values during initialisation of the appender. This release introduces support for ctx (MDC) parameter lookup done in a dynamic way - per log message.

This functionality uses internal log4j2 classes so logic is the same as in Lo4j2. The implementation is not optimal and allocates objects for each message. This will be addressed in a future release.

0.9.9 Support for Java 8

Whole project is configured to compile for and target java 8.

0.9.8 Support for Java 8

Whole project is configured to compile for and target java 8.

0.9.7 Support for Java 8

Whole project is configured to compile for and target java 8.

0.9.6 Support for Grafana cloud, HTTPS and basic auth

Log4j appender:

• Added support for https (when port is configured to be 443).
• Added support for basic auth with username and password specified in configuration.
• Added test that connects to Grafana Cloud account and verifies that integration works.

Logback appender:

• Added efficient pattern layout that allocates less.

0.9.5 New Logback appender!

Tjahzi got some minor updates, code cleanups and fixes, but also 0.1 release of Logback appender.

1. Replaced TextBuilder from javolution with our own version to reduce dependencies.
2. Fixed bug causing Tjahzi to complain about wrong buffer size when it is in fact correct.
3. Released 0.1 version of Logback appender.

0.9.4 Changed namespace and proper no dependency release

Changed artifacts namespace to pl.tkowalcz.tjahzi. Released proper "nodep" artifact.

0.9.3 Working on renaming modules and no dependency release

Please disregard this release.

0.9 Custom protobuf serializers

In this release a lot of work has gone into reducing allocations. We no longer need to materialize any strings from log buffer and all data is serialized garbage free to protobuf format.

Ticket #5 removed from project protobuf generated by official toolchain. Now we are generating protobuf messages by hand with dedicated code (which is quite simple really). This has also substantial impact on the size of dependnecies. The protobuf jar was over 800kb in size. It is still generated and used for testing.

There were other optimisations in area of object allocation. Added test that use com.google.code.java-allocation-instrumenter to inspect allocations.