OTP R14A今天发布了

以下是这次发布的亮点，没有太大的性能改进， 主要是修理了很多BUG！

--- HIGHLIGHTS ----------------------------------------------------------

    OTP-8217  == erts stdlib compiler ==

      The module binary from EEP31 (and EEP9) is implemented.

    OTP-8485  == common_test ==

      It is now possible for the user to provide specific callback
      modules that handle test configuration data, so that data on
      arbitray form can be accessed (e.g. by reading files or by
      communicating with a configuration server process). Two
      default callback modules have been introduced in Common Test:
      ct_config_plain and ct_config_xml. The former is used to
      handle the traditional Common Test configuration files (with
      terms on key-value tuple form) and the latter to handle
      configuration data on XML representation.

    OTP-8555  == erts ==

      New NIF features:

      Send messages from a NIF, or from thread created by NIF, to
      any local process (enif_send)

      Store terms between NIF calls (enif_alloc_env,
      enif_make_copy)

      Create binary terms with user defined memory management
      (enif_make_resource_binary)

      And some incompatible changes made to the API. For more
      information see the warning text in erl_nif(3).

    OTP-8623  == compiler erts hipe stdlib ==

      Receive statements that can only read out a newly created
      reference are now specially optimized so that it will execute
      in constant time regardless of the number of messages in the
      receive queue for the process. That optimization will benefit
      calls to gen_server:call(). (See gen:do_call/4 for an example
      of a receive statement that will be optimized.)

    OTP-8650  == common_test ==

      The run_test script has been replaced by a program (with the
      same name) which can be executed without explicit
      installation. The start flags are the same as for the legacy
      start script.

    OTP-8706  == erts hipe tools ==

      eprof has been reimplemented with support in the Erlang
      virtual machine and is now both faster (i.e. slows down the
      code being measured less) and scales much better. In
      measurements we saw speed-ups compared to the old eprof
      ranging from 6 times (for sequential code that only uses one
      scheduler/core) up to 84 times (for parallel code that uses 8
      cores).

      Note: The API for the eprof has been cleaned up and extended.
      See the documentation.

期待R14B有大的动作！