Skip to content

9.37 Release

Compare
Choose a tag to compare
@jfriesne jfriesne released this 03 May 01:27
· 146 commits to master since this release

9.37 - Released 5/2/2024

  • Implemented the ustar extension in TarFileWriter to support file paths
    and linked-file paths up to 256 characters in length.
  • Implemented the pax extension in TarFileWriter to support file paths
    and linked-file paths of unlimited length.
  • Added CPut*() convenience methods to the Python Message class.
  • Updated the snoopsharedmem to use more sophisticated argument parsing.
  • Updated the snoopsharedmem to have "delay=N" and "clear" options.
  • Added a GetThreadPriorityName() convenience-method to the Thread class.
  • Added a MUSCLE_AVOID_THREAD_PRIORITIES flag that can be added to the
    compile-line to turn all set-thread-priority calls into no-ops.
  • Added SetThreadScheduler()/GetThreadScheduler()/GetThreadSchedulerName()
    methods to the Thread class to support Linux's SCHED_FIFO, etc.
  • Updated the Python2 and Python3 message.py classes so that GetFieldItem()
    and the Get*() methods now take an optional (defaultValue) argument that
    specifies what should be returned if the requested item is not present
    in the Message.
  • Removed the option for MUSCLE_NODISCARD macro to expand to
    __attribute__((warn_unused_result)) under pre-C++17 codebases,
    as that implementation warns even when an explicit (void) cast
    is employed, causing too much visual noise to be useful.
  • Replaced calls to newnothrow with calls to new, to simplify the
    code given fact that in modern OS's newnothrow's successful return
    doesn't actually mean the RAM is available anyway.
  • Added RoundUp() versions of all the TimeUnitConversion functions
    (e.g. added MicrosToMillisRoundUp() to complement MicrosToMillis())
  • Modified the MUSCLE_USE_POLL and MUSCLE_USE_EPOLL implementations of
    SocketMultiplexer to round up to the nearest millisecond when computing
    timeout-times (instead of rounding down, since slightly-early-wakeups
    are worse than slightly-late-wakeups)