4mc 3.0.0

Namespace root and group id renamed to com.fing.* as preliminary step to maven-central publishing.
Native libraries have been updated to LZ4 1.9.4 and ZSTD 1.5.2.
Recompiled all native libraries and tools.

4mc 2.2.0

Native libraries have been updated to LZ4 1.9.2 and ZSTD 1.4.4.
Recompiled all native libraries and tools.

4mc 2.1.0

Compatibility with newer Hadoop (2.7.x) and Spark (2.4.3).

4mc 2.0.0

ZSTD compression library (https://github.com/facebook/zstd) has been integrated and releases with a new file format (4mz) and codecs (FourMz family).

4mc 1.4.0

• Embedded native libraries (like jpountz-lz4) to avoid need of separate lib and make it easily runnable on Hadoop, Spark, Flink, Storm jobs, both on local env (e.g. on Windows/Mac) and on remote env (Linux).
• Bug fixes of compression corner cases.
• Compiled tool '4mc' available at source baseline in tool folder.

4mc 1.3.0

Introduced direct memory buffer pools to cope with "java.lang.OufOfMemoryError: Direct Buffer Memory".
This is not happening on cluster jobs but if client apps open huge number of 4mc files, it happens with lazy GC that direct nio buffers are alive and their dedicated memory go over.
With dedicated cache pool of these buffers client applications are much safer, but in worst case it's still not enough, the recommended setting is to increase direct buffer memory of your app by e.g.:
-XX:MaxDirectMemorySize=512m

4mc 1.2.0

Minor fixes on class visibility and stream closure in index block reading.

4mc 1.1.0

Support hadoop-1 and hadoop-2 seamlessly

4mc 1.0.0

Very first version of 4mc.
Released artifacts contain java compiled library and 4mc.exe tool for Windows.
4mc CLI Linux/MacOs and JNI library: better you compile it for your platofrm using provided Makefile