ruby--ruby/NEWS

# -*- rdoc -*-

= NEWS for Ruby 2.2.0

This document is a list of user visible feature changes made between
releases except for bug fixes.

Note that each entry is kept so brief that no reason behind or
reference information is supplied with.  For a full list of changes
with all sufficient information, see the ChangeLog file.

== Changes since the 2.1.0 release

=== Language changes

=== Core classes updates (outstanding ones only)

* Enumerable
  * New methods:
    * Enumerable#slice_after
  * Extended methods:
    * min, min_by, max and max_by supports optional argument to return
      multiple elements.

* Float
  * New methods:
    * Float#next_float
    * Float#prev_float

* File
  * New methods:
    * File.birthtime
    * File#birthtime

* File::Stat
  * New methods:
    * File::Stat#birthtime

* Process
  * Extended method:
    * Process execution methods such as Process.spawn opens the file in write
      mode for redirect from [:out, :err].

* Symbol
  * New methods
    * Symbol.find(str) returns whether given string is defined as symbol or not.
  * Improvements
    * Most symbols which are returned by String#to_sym and
      String#intern are GC-able.

* Matrix
  * New methods:
    * Matrix#first_minor(row, column) returns the submatrix obtained
      by deleting the specified row and column.
    * Matrix#cofactor(row, column) returns the (row, column) cofactor
      which is obtained by multiplying the first minor by (-1)**(row + column).

=== Core classes compatibility issues (excluding feature bug fixes)

* IO
  * incompatible changes:
    * When flushing file IO, you cannot assume that the metadata of the file
      is updated immediately.  On some platforms (especially Windows), it is
      delayed until the filesystem load is decreased.

* Math
  * incompatible changes:
    * Math.log now raises Math::DomainError instead of returning NaN if the
      base is less than 0, and returns NaN instead of -infinity if both of
      two arguments are 0.
    * Math.atan2 now returns values like as expected by C99 if both two
      arguments are infinity.

* Proc
  * incompatible changes:
    * ArgumentError is no longer raised when lambda Proc is passed as a
      block, and the number of yielded arguments does not match the formal
      arguments of the lambda, if just an array is yielded and its length
      matches.

* Process
  * Process execution methods such as Process.spawn opens the file in write
    mode for redirect from [:out, :err].
    Before Ruby 2.2, it was opened in read mode.

=== Stdlib updates (outstanding ones only)

* Etc
  * New methods:
    * Etc.uname
    * Etc.sysconf
    * Etc.confstr
    * IO#pathconf

* Find, Pathname
  * Extended methods:
    * find method accepts "ignore_error" keyword argument.

* Pathname
  * Pathname#/ is aliased to Pathname#+.

=== Stdlib compatibility issues (excluding feature bug fixes)

* time.rb
  * Time.parse, Time.strptime, Time.rfc2822, Time.xmlschema may produce
    fixed-offset Time objects.
    It is happen when usual localtime doesn't preserve the offset from UTC.
  * Time.httpdate produces always UTC Time object.
  * Time.strptime raises ArgumentError when no date information.

* lib/test/**/*.rb
  * Removed because it conflicts to minitest 5, and it was just an wrapper
    of minitest 4. [Feature #9711]

* Prime
  * incompatible changes:
    * Prime.prime? now returns false for negative numbers. This method
      should not be used to know the number is composite or not. [Bug #7395]

=== Built-in global variables compatibility issues

=== C API updates

* Deprecated APIs removed.  [Feature #9502]

  Check_SafeStr -> SafeStringValue
  rb_check_safe_str -> SafeStringValue
  rb_quad_pack -> rb_integer_pack
  rb_quad_unpack -> rb_integer_unpack
  rb_read_check : access struct FILE internal. no replacement.
  rb_struct_iv_get : internal function. no replacement.
  struct rb_blocking_region_buffer : internal type. no replacement.
  rb_thread_blocking_region_begin -> rb_thread_call_without_gvl family
  rb_thread_blocking_region_end -> rb_thread_call_without_gvl family
  TRAP_BEG -> rb_thread_call_without_gvl family
  TRAP_END -> rb_thread_call_without_gvl family
  rb_thread_select -> rb_thread_fd_select
  struct rb_exec_arg : internal type. no replacement.
  rb_exec : internal function. no replacement.
  rb_exec_arg_addopt : internal function. no replacement.
  rb_exec_arg_fixup : internal function. no replacement.
  rb_exec_arg_init : internal function. no replacement.
  rb_exec_err : internal function. no replacement.
  rb_fork : internal function. no replacement.
  rb_fork_err : internal function. no replacement.
  rb_proc_exec_n : internal function. no replacement.
  rb_run_exec_options : internal function. no replacement.
  rb_run_exec_options_err : internal function. no replacement.
  rb_thread_blocking_region -> rb_thread_call_without_gvl family
  rb_thread_polling -> rb_thread_wait_for
  rb_big2str0 : internal function. no replacement.
  rb_big2ulong_pack -> rb_integer_pack
  rb_gc_set_params : internal function. no replacement.

* struct RBignum is hidden.  [Feature #6083]
  Use rb_integer_pack and rb_integer_unpack instead.

* struct RRational is hidden.  [Feature #9513]
  Use rb_rational_num and rb_rational_den instead.

* rb_big_new and rb_big_resize takes a size_t instead of long.

* rb_num2long returns a long instead of SIGNED_VALUE.

* rb_num2ulong returns an unsigned long instead of VALUE.

* st hash table uses power-of-two sizes for speed [Feature #9425].
  Lookups are 10-25% faster if using appropriate hash functions.
  However, weaknesses in hash distribution can no longer be masked
  by prime number-sized tables, so extensions may need to tweak
  hash functions to ensure good distribution.

* rb_sym2str() added. This is almost same as `rb_id2str(SYM2ID(sym))`
  but not pinning a dynamic symbol.

* struct RSymbol added. This represents a dynamic symbol as object in
  Ruby's heaps.

* rb_str_cat_cstr() added. This is same as `rb_str_cat2()`.

* `rb_str_substr()` and `rb_str_subseq()` now share middle of a string,
  but not only the end of a string.  Therefore, result strings may not
  be NUL-terminated, `StringValueCStr()` is needed calling to obtain a
  NUL-terminated C string.