v1.16.0

v1.16.0 brings a rewrite of Maxwell's filtering system, giving it a
concise list of rules that are executed in sequence. It's now possible
to exclude tables from a particular database, exclude columns matching a
value, and probably some other use cases.
See http://maxwells-daemon.io/config/#filtering for details.

v1.15.0

This is a bug-fix release, but it's big enough I'm giving it a minor
version.

• Fix a very old bug in which DDL rows were writing the start of the
  row into maxwell.positions, leading to chaos in some scenarios where
  maxwell managed to stop on the row and double-process it, as well as to
  a few well-meaning patches.
• Fix the fact that maxwell was outputting "next-position" instead of
  "position" of a row into JSON.
• Fix the master-recovery code to store schema that corresponds to the
  start of a row, and points the replicator at the next-position.

Much thanks to Tim, Likun and others in sorting this mess out.

v1.14.7

• add RowMap#getRowQuery, thx @saimon7
• revert alpine-linux docker image fiasco
• fix RawJSONString not serializable, thx @niuhaifeng

v1.14.6

• Fix docker image

v1.14.5

• reduce docker image footprint
• add benchmarking framework
• performance improvements for date/datetime columns
• fix parser error on UPGRADE PARTITIONING

v1.14.4

• Fix race condition in SchemaCapturer

v1.14.3

• Enable jvm metrics

v1.14.2

• fix regression in 1.14.1 around bootstrapping host detection
• fix heartbeating code around table includes

v1.14.1

• bootstraps can now take a client_id
• improved config validation for embedded mode

v1.14.0

• new feature --output_xoffset to uniquely identify rows within transactions,
  thx Jens Gyti
• Bug fixes around "0000-00-00" times.
• Bug fixes around dates pre 1000 AD