pike.git / CHANGES

pike.git/CHANGES:7:   o Crypto.Hash.SCRAM       New class to support SCRAM authentication.      o Concurrent       - Add map_with() as an alias to flat_map().       - Add a convenience variant to zip().    -  - depend() can be used to collect multiple dependencies before finalising -  the Promise. +  - depend() can be used to collect multiple dependencies before +  finalising the Promise.      o Debug.find_all_clones()    -  New convenience function to find all objects that are clones -  of a program. +  New convenience function to find all objects that are clones of a +  program.      o MIME       Added a setter for the message boundary prefix.      o Sql.pgsql       - Added support for PostgreSQL 10's stronger password authentication    (SCRAM-SHA-256).   
pike.git/CHANGES:72:    Support the new terminfo format from ncurses 6.      o Thread.Farm       Added Thread.Farm()->set_thread_name_cb() to help applications    that monitor thread creation and termination for the purpose    of tracking thread names.      o Thread.ResourceCount    -  New module to allow threadsafe race-condition-free counting of resources. +  New module to allow threadsafe race-condition-free counting of +  resources.      o Tools.Standalone.check_http       - Allow redirect responses.       - Support --expect option.      o Web.Sass       This is a module that interfaces to the SASS compilation    library "LibSass" (https://github.com/sass/libsass).         Bug fixes   ---------      o ADT.CritBit    -  - Due to an off-by-one error the last limb of bignums was never compared. -  This resulted (sometimes) in unequal bignums comparing equal inside of -  the tree. +  - Due to an off-by-one error the last limb of bignums was never +  compared. This resulted (sometimes) in unequal bignums comparing +  equal inside of the tree.      o Compiler       - Improved variant robustness.       - Fixed type leak for getters/setters implemented with    variant functions.      o Calendar       - Updated tzdata to 2018d.       - Calendar.TimeRanges.NullTimeRange now has a working __hash().    The expression ([ Calendar.TimeRanges.nulltimerange : 17 ])    no longer gives a backtrace.      o Concurrent    -  - Less filling, deadlock-free, faster, available without running backend. +  - Less filling, deadlock-free, faster, available without running +  backend.       - Handle an empty argument array properly in results().       Previously code such as:       Concurrent.Future f2 = Concurrent.results(({}));    -  f2->on_success(lambda(array(string) a) { werror("success: %O.\n", a); }) -  ->on_failure(lambda(mixed err) { werror (describe_backtrace(err)); }); +  f2->on_success(lambda(array(string) a) { +  werror("success: %O.\n", a); +  }) +  ->on_failure(lambda(mixed err) { +  werror (describe_backtrace(err)); +  });    -  would lead to a backtrace because the future was destructed prematurely. +  would lead to a backtrace because the future was destructed +  prematurely.       - Readable/understandable documentation.      o Crypto.CCM       Fixed initialization of the iv.      o Crypto.Hash       Censor the password argument from backtraces in crypt_hash().      o Filesystem.Monitor       - Fixed some backend_check rescheduling isusues.    -  adjust_monitor() didn't reschedule the backend_check() call_out -  if the adjusted monitor already was at the head of monitor_queue. +  adjust_monitor() didn't reschedule the backend_check() call_out if +  the adjusted monitor already was at the head of monitor_queue.    This caused polls to be delayed in some common cases.       register_path() didn't reschedule the backend_check() call_out. -  If the newly registered monitor ends up at the head of the monitor_queue, -  the backend_check() call_out likely needs to be rescheduled. +  If the newly registered monitor ends up at the head of the +  monitor_queue, the backend_check() call_out likely needs to be +  rescheduled.    -  - Fixed issue with next_poll being set to zero causing polling -  every second. +  - Fixed issue with next_poll being set to zero causing polling every +  second.    -  - Fixed indexing the NULL value error when InotifyMonitor was used in -  combination with a filter_file() that doesn't accept everything. +  - Fixed indexing the NULL value error when InotifyMonitor was used +  in combination with a filter_file() that doesn't accept +  everything.       - Fixed incorrect creation of new monitors.       Monitors were created using monitor() on the top level rather than    using Monitor::monitor(), which prevented the symlinks' overridden -  DefaultMonitor from setting the correct state on newly created monitors. +  DefaultMonitor from setting the correct state on newly created +  monitors.       We now force a check on the directory monitor instead, which will    pick up the new file instantly and create the sub monitor.       - Fixed race condition on deletion.    -  This fixes a race that could occur on rapid exists => delete => exists -  transitions, where the directory monitor never noticed the change but -  the sub monitor was removed from the monitor lookup mapping. The -  directory monitor is now notified on sub-monitor release. +  This fixes a race that could occur on rapid exists => delete => +  exists transitions, where the directory monitor never noticed the +  change but the sub monitor was removed from the monitor lookup +  mapping. The directory monitor is now notified on sub-monitor +  release.       - Added set_stable_time() function.       - Removed obsolete special case for co_id == 1.      o HTTPLoop:       Fixed some potential race conditions.      o Image.Image    -  - Improved argument checking in apply_curve(). -  Non-integer array values lead to use of uninitialized curve values. +  - Improved argument checking in apply_curve(). Non-integer array +  values lead to use of uninitialized curve values.      o Image.PVR       Fixed size check for twiddled images.      o MIME       Support trailing newline in MIME part epilogues.      o Protocols.HTTP.Query       - Fixed incoming chunked transfers (timed async works now, chunk    options are properly 'parsed'). Trailer headers are now correctly    parsed.       - Improved handling of multiple headers of the same kind.    -  - Fixed a bug where base64 encoded HTTP basic auth tokens were encoded in a -  pretty-printed format including line breaks, which lead to malformed HTTP -  requests on long username and password combinations. +  - Fixed a bug where base64 encoded HTTP basic auth tokens were +  encoded in a pretty-printed format including line breaks, which +  lead to malformed HTTP requests on long username and password +  combinations.       - Do not wait for SSL.File()->close() to complete when called    from the destruct() callback.      o Protocols.HTTP.Server       - The header parser exception mode now doesn't immediately throw an    exception.       - Attempt to ensure that data is sent before terminating an
pike.git/CHANGES:231:      o Search.Utils       Updated debug string format.      o Sql.pgsql       - Suppress spurious 'File not open' blurts on stderr when the    connection is still pending and we inspect the connection object.    -  - Signal EOF early before closing the SQL result portal and thoroughly -  cleanup bufcon/stashcount to prevent occasional deadlocks -  on asynchronous object destructions. +  - Signal EOF early before closing the SQL result portal and +  thoroughly cleanup bufcon/stashcount to prevent occasional +  deadlocks on asynchronous object destructions.       - Suppress rare "access in destructed object" messages.    -  - Simplified fundamentally flawed reconnect logic, made failures consistent. +  - Simplified fundamentally flawed reconnect logic, made failures +  consistent.       - Repair SSL connect logic.       - Update documentation.       - Repair fetch_row_array().       - Ensure that all database errors result in exceptions (eventually).       - affected_rows() and status_command_complete() now wait until they
pike.git/CHANGES:272:    - Fix memory leak in strftime().      o Standards.IIM       Disabled debug output for unknown segment markers.      o Stdio       - Restore the FD_SETSIZE on NT to 65536.    -  Due to #include file restructuring, the FD_SETSIZE value on -  NT had fallen back to the default (64). This issue has probably -  affected all releases of Pike since 7.9.2. +  Due to #include file restructuring, the FD_SETSIZE value on NT had +  fallen back to the default (64). This issue has probably affected +  all releases of Pike since 7.9.2.       - mv() [NT]       Fixed single byte buffer overrun in mv() on NT.       A NUL-terminator was written outside a malloced buffer. This    caused intermittent crashes after unusual use of mv().       Fixes the crash in [PIKE-90].   
pike.git/CHANGES:335:    Support MariaDB 10.2 headerfiles.      o export       - Fixed fix_configure.       This was broken in multiple ways:       * Did not consider timestamp on aclocal.m4    * Did not consider post_modules -  * Did not check configure scripts directly under modules or post_modules -  * When fixing modules, it looked for the configure scripts in the build -  directory rather than the source directory +  * Did not check configure scripts directly under modules or +  post_modules +  * When fixing modules, it looked for the configure scripts in the +  build directory rather than the source directory    * Did not use the run_autoconfig wrapper       This has caused xenofarm to export snapshots with out of date    configure scripts.      o precompile       Support precompilation with Pike 8.1 again.      o run_autoconfig
pike.git/CHANGES:361:   o smartlink       Fixed multiple issues on Darwin & NetBSD.      o Whitefish       Use the standard VPATH.      o Yp    -  Support using libtirpc. Glibc 2.26 has removed the Sun RPC implementation... +  Support using libtirpc. Glibc 2.26 has removed the Sun RPC +  implementation.         Changes since Pike 8.0.466 (release 10)   ----------------------------------------------------------------------      New Features   ------------      o Crypto.ECC