[ntp:commitlogs] incoming pull in deacon.udel.edu:/deacon/backroom/ntp-dev

Harlan Stenn stenn at deacon.udel.edu
Wed Mar 2 00:05:13 UTC 2011


VAR       CLIENT              SERVER              
===       ======              ======              
USER      stenn               stenn               
HOST      deacon.udel.edu     psp-deb1.ntp.org    
ROOT      /deacon/backroom/ntp-dev/home/hart/ntp-dev-ev-cached-tod
LEVEL     1                   1                   
TIME_T    1297863471          1297863471          
UTC       20110216133751      20110216133751      
VERSION   bk-5.1.1            bk-5.1.1            

stenn at deacon.udel.edu fired the post-incoming--012commitlogs trigger in /deacon/backroom/ntp-dev
    Received the following changesets
    
    ChangeSet at 1.2487, 2011-03-01 04:48:43+00:00, davehart at shiny.ad.hartbrothers.com
      libevent: When building on systems with CLOCK_MONOTONIC available,
        separate the internal timeline (possibly counting since system boot)
        from the gettimeofday() timeline in event_base cached timevals.  Adds
        new event_base_tv_cached() to retrieve cached callback round start
        time on the internal timeline, and changes
        event_based_gettimeofday_cached() to always return times using the
        namesake timeline.  This preserves the benefit of using the never-
        stepped monotonic clock for event timeouts while providing clients
        with times consistently using gettimeofday().
      Correct event_base_gettimeofday_cached() workaround code in
        sntp to work with corrected libevent.
      Remove sntp l_fp_output() test now that it uses prettydate().
    
      BitKeeper/deleted/9f/debug-input-lfp-hex~523a3dc59a6a4e86 at 1.3 +0 -0
        Rename: sntp/tests/data/debug-input-lfp-hex -> BitKeeper/deleted/9f/debug-input-lfp-hex~523a3dc59a6a4e86
    
      ChangeLog at 1.804 +12 -0
        libevent: When building on systems with CLOCK_MONOTONIC available,
          separate the internal timeline (possibly counting since system boot)
          from the gettimeofday() timeline in event_base cached timevals.  Adds
          new event_base_tv_cached() to retrieve cached callback round start
          time on the internal timeline, and changes
          event_based_gettimeofday_cached() to always return times using the
          namesake timeline.  This preserves the benefit of using the never-
          stepped monotonic clock for event timeouts while providing clients
          with times consistently using gettimeofday().
        Correct event_base_gettimeofday_cached() workaround code in
          sntp to work with corrected libevent.
        Remove sntp l_fp_output() test now that it uses prettydate().
    
      libntp/socket.c at 1.6 +3 -6
        Add \n termination overlooked in msyslog()->DPRINTF() changes
    
      libntp/work_fork.c at 1.8 +1 -1
        Add \n termination overlooked in msyslog()->DPRINTF() changes
    
      sntp/libevent/configure.in at 1.3 +4 -0
        Add AM_SILENT_RULES patch now in git libevent.
    
      sntp/libevent/event-internal.h at 1.2 +8 -1
        Add struct timeval tod_tv_cache to struct event_base, analogous
          to tv_cache but always using the gettimeofdday() timeline.  Not
          needed on systems without CLOCK_MONOTONIC.
    
      sntp/libevent/event.c at 1.2 +73 -5
        When building on systems with CLOCK_MONOTONIC available, separate the
          internal timeline (possibly counting since system boot) from the
          gettimeofday() timeline in event_base cached timevals.  Adds new
          event_base_tv_cached() to retrieve cached callback round start time
          on the internal timeline, and changes
          event_based_gettimeofday_cached() to always return times using the
          namesake timeline.  This preserves the benefit of using the never-
          stepped monotonic clock for event timeouts while providing clients
          with times consistently using gettimeofday().
    
      sntp/libevent/evthread.c at 1.2 +11 -1
        Add signature to struct debug_lock so that invalid lock pointers
          can be caught sooner.
    
      sntp/libevent/include/event2/event.h at 1.2 +18 -3
        Add event_base_tv_cached() to retrieve the start-of-callbacks
          time on the internal timeline (possibly monotonic).
    
      sntp/m4/ntp_libevent.m4 at 1.6 +2 -1
        Respect a hypothetical -R/some/lib/path in libevent.pc.
        Pass --enable-silent-rules to tearoff libevent configure.
    
      sntp/m4/ntp_openssl.m4 at 1.17 +1 -1
        Correct comment to match code.
    
      sntp/main.c at 1.69 +121 -49
        Use pthread locks/mutexes/cond vars despite our so-far single-threaded
          use of libevent.
        Correct event_base_gettimeofday_cached() workaround code in
          sntp to work with corrected libevent.
    
      sntp/networking.c at 1.58 +31 -25
        Show server address in wrong org timestamp message.
    
      sntp/tests/Makefile.am at 1.24 +0 -2
        Remove sntp l_fp_output() test now that it uses prettydate().
    
      sntp/tests/packetProcessing.cpp at 1.5 +4 -4
        whitespace cleanup only
    
      sntp/tests/utilities.cpp at 1.7 +0 -16
        Remove sntp l_fp_output() test now that it uses prettydate().
    
      sntp/utilities.c at 1.17 +5 -14
        use prettydate() for l_fp_output()
    
    ChangeSet at 1.2486, 2011-02-25 12:37:40+00:00, davehart at shiny.ad.hartbrothers.com
      [Bug 1839] 4.2.7p131 installs libevent ev*.h headers.
    
      ChangeLog at 1.803 +1 -0
        [Bug 1839] 4.2.7p131 installs libevent ev*.h headers.
    
      sntp/libevent/include/Makefile.am at 1.2 +9 -2
        [Bug 1839] 4.2.7p131 installs libevent ev*.h headers.
    
    ChangeSet at 1.2485, 2011-02-25 12:36:58+00:00, davehart at shiny.ad.hartbrothers.com
      Ensure CONFIG_SHELL is not empty before relying on it for #! scripts.
    
      ChangeLog at 1.802 +1 -0
        Ensure CONFIG_SHELL is not empty before relying on it for #! scripts.
    
      configure.ac at 1.519 +1 -0
        Ensure CONFIG_SHELL is not empty before relying on it for #! scripts.
    


More information about the commitlogs mailing list