# -*- rd -*-
= NEWS

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 1.9.2 release
=== License

* Ruby's License is changed from a dual license with GPLv2
  to a dual license with 2-clause BSDL.

=== Known platform dependent issues
==== OS X Lion

* You have to configure ruby with '--with-gcc=gcc-4.2' if you're using
  Xcode 4.1, or, if you're using Xcode 4.2, you have to configure ruby
  with '--with-gcc=clang'.

=== C API updates

* rb_scan_args() is enhanced with support for option hash argument
  extraction.

* ruby_vm_at_exit() added.  This enables extension libs to hook a VM
  termination.

* rb_reserved_fd_p() added.  If you want to close all file descriptors,
  check using this API.  [ruby-core:37759]

=== Library updates (outstanding ones only)

* builtin classes

  * ARGF
    * new methods:
      * ARGF.print
      * ARGF.printf
      * ARGF.putc
      * ARGF.puts
      * ARGF.read_nonblock
      * ARGF.to_write_io
      * ARGF.write

  * Array
    * extended method:
      * Array#pack supports endian modifiers

  * Bignum
    * Multiplication algorithm for Bignums with a large number of digits over
      150 BDIGITs is changed in order to reduce its calculation time.
      Now such large Bignums are multiplied by using Toom-3 algorithm.

  * Encoding
    * new encodings:
      * CP950
      * CP951
      * UTF-16
      * UTF-32
    * change alias:
      * SJIS is Windows-31J

  * File
    * new constant:
      * File::NULL
        name of NULL device.
      * File::DIRECT
        name of O_DIRECT.

  * IO
    * extended method:
      * IO#putc supports multibyte characters
    * new methods:
      * IO#advise
      * IO.write(name, string, [offset] )
        Write `string` to file `name`.
        Opposite with File.read.
      * IO.binwrite(name, string, [offset] )
        binary version of IO.write.

  * Kernel
    * move #__id__ to BasicObject.
    * extended method:
      * Kernel#rand supports range argument

  * Module
    * new methods:
      * Module#private_constant
      * Module#public_constant

  * Random
    * extended method:
      * Random.rand supports range argument

  * String
    * extended method:
      * String#unpack supports endian modifiers
    * new method:
      * String#prepend
      * String#byteslice

  * Time
    * extended method:
      * Time#strftime supports %:z and %::z.

  * Process
    * Process#maxgroups and Process#maxgroups= now raise NotImplementedError if
      the platform don't support supplementary groups concept.

* bigdecimal

  * BigDecimal#power and BigDecimal#** support non-integral exponent.

  * Kernel.BigDecimal and BigDecimal.new now accept instances of Integer,
    Rational, Float, and BigDecimal.  If you pass a Rational or a Float to
    them, you must specify the precision to produce the digits of a BigDecimal.

  * The behavior of BigDecimal#coerce with a Rational is changed.  It uses
    the precision of the receiver BigDecimal to produce the digits of a
    BigDecimal from the given Rational.

* bigdecimal/util

  * BigDecimal#to_d and Integer#to_d are added.

  * Float#to_d accepts a precision.

  * Rational#to_d raises ArgumentError when passing zero or negative
    precision.

  * Rational#to_d

    * Zero and an implicit precision is deprecated.
      This feature is removed at the next release of bigdecimal.

    * A negative precision isn't supported.
      Be careful it is an incompatible change.

* date

  * Accepts flonum explicitly with limitations.
    * If the given offset is flonum, DateTime assumes its precision is
      at most second.

      DateTime.new(2001,2,3,0,0,0,3.0/24) ==
      DateTime.new(2001,2,3,0,0,0,'+03:00')
        #=> true

    * If the given operand for -/+ is flonum, DateTime assumes its
      precision is at most nanosecond.

      DateTime.new(2001,2,3) + 0.5 == DateTime.new(2001,2,3,12)
        #=> true

    * Precision of offset is always at most second.

      Rational('0.5') == Rational('0.500001')  #=> false
      DateTime.new(2001,2,3,0,0,0,Rational('0.5')) ==
      DateTime.new(2001,2,3,0,0,0,Rational('0.500001'))
        #=> true

  * Ignores long offset and far reform day (with warning).

    * Now accepts only:

      -1<=offset<=1 (-24:00..+24:00)
      2298874<=start<=2426355 or -/+oo
        (proleptic Gregorian/Julian mean -/+oo)

  * A method strftime cannot produce huge output (same as Time's one).

    * Even though Date/DateTime can handle far dates, the following causes
      an exception.

      DateTime.new(1<<10000).strftime('%Y')  # Errno::ERANGE

  * Changed the format of inspect.
  * Changed the format of marshal (but, can load old dumps).

* io/console
  * new methods:
    * IO#noecho {|io| }
    * IO#echo=
    * IO#echo?
    * IO#raw {|io| }
    * IO#raw!
    * IO#getch
    * IO#winsize
    * IO.console

* json
  * updated to v1.5.4.

* matrix
  * new classes:
    * Matrix::EigenvalueDecomposition
    * Matrix::LUPDecomposition
  * new methods:
    * Matrix#diagonal?
    * Matrix#eigen
    * Matrix#eigensystem
    * Matrix#hermitian?
    * Matrix#lower_triangular?
    * Matrix#lup
    * Matrix#lup_decomposition
    * Matrix#normal?
    * Matrix#orthogonal?
    * Matrix#permutation?
    * Matrix#round
    * Matrix#symmetric?
    * Matrix#unitary?
    * Matrix#upper_triangular?
    * Matrix#zero?
    * Vector#magnitude, #norm
    * Vector#normalize
  * extended methods:
    * Matrix#each and #each_with_index can iterate on a subset of the elements
    * Matrix#find_index returns [row, column] and can iterate on a subset
      of the elements
    * Matrix#** implements Numeric exponents (using the eigensystem)
    * Matrix.zero can build rectangular matrices

* minitest
  * Minitest has been updated to version 2.2.2.
  * For full details, see https://github.com/seattlerb/minitest/blob/master/History.txt

* net/http
  * SNI (Server Name Indication) supported for HTTPS.

  * Allow to configure to wait server returning '100 continue' response
    before sending HTTP request body. Set Net::HTTP#continue_timeout AND pass
    'expect' => '100-continue' to a extra HTTP header.

    For example, the following code sends HTTP header and waits for getting
    '100 continue' response before sending HTTP request body. When 0.5 [sec]
    timeout occurs or the server send '100 continue', the client sends HTTP
    request body.
      http.continue_timeout = 0.5
      http.request_post('/continue', 'body=BODY', 'expect' => '100-continue')

  * new method:
    * Net::HTTPRequest#set_form): Added to support
      both application/x-www-form-urlencoded and multipart/form-data.

* objspace
  * new method:
    * ObjectSpace::memsize_of_all

* openssl
  * PKey::RSA and PKey::DSA now use the generic X.509 encoding scheme
    (e.g. used in a X.509 certificate's Subject Public Key Info) when
    exporting public keys to DER or PEM. Backward compatibility is
    ensured by (already existing) fallbacks during creation.
  * OpenSSL::ASN1::Constructive#new and OpenSSL::ASN1::Primitive#new
    (and the constructors of their sub-classes) will no longer force
    tagging to be set to :EXPLICIT when tag and/or tag_class are passed
    as parameters. tagging must be set explicitly.
  * Support for infinite length encodings via infinite_length attribute.
  * OpenSSL::PKey.read( file | string [, pwd] ) allows to read arbitrary
    public/private keys in DER-/PEM-encoded form with an optional password
    for encrypted PEM encodings.
  * Add new method OpenSSL::X509::Name#hash_old as a wrapper of
    X509_NAME_hash_old() defined from OpenSSL 1.0.0. It returns OpenSSL 0.9.8
    compatible hash value.

* optparse
  * support for bash/zsh completion.

* Rake
  * Rake has been upgraded from 0.8.7 to 0.9.2.2.  For full release notes see
    https://github.com/jimweirich/rake/blob/master/CHANGES

* RDoc
  * RDoc has been upgraded to version 3.9.4.  For full release notes see
    http://docs.seattlerb.org/rdoc/History_txt.html

* rexml
  * Support Ruby native encoding mechanism and iconv dependency is dropped.

* RubyGems
  * RubyGems has been upgraded to version 1.8.10. For full release notes see
    http://rubygems.rubyforge.org/rubygems-update/History_txt.html

* stringio
  * extended method:
    * StringIO#set_encoding can get 2nd argument and optional hash.

* test/unit
  * New arguments:
    * -j N, --jobs=N: Allow run N testcases at once.
    * --jobs-status: Show status of jobs when parallel running.
    * --no-retry: Don't retry testcases which failed when parallel running.
    * --ruby=RUBY: path to ruby for job(worker) process. optional.
    * --hide-skip: Hide skip messages. You'll see the number of skips at end of
      test result.

* uri
  * new methods:
    * URI::Generic#hostname
    * URI::Generic#hostname=

* webrick
  * new method:
    * WEBrick::HTTPRequest#continue for generating '100 continue' response.
  * new logging directive:
    * %{remote}p for remote (client) port number.

* yaml
  * The default YAML engine is now Psych. You may downgrade to syck by setting
    YAML::ENGINE.yamler = 'syck'.

* zlib
  * new methods:
    * Zlib.deflate
    * Zlib.inflate

* FileUtils
  * extended method:
    * FileUtils#chmod supports symbolic mode argument.

=== Language changes

* Regexps now support Unicode 6.0. (new characters and scripts)

* [experimental] Regexps now support Age property.
  Unlike Perl, current implementation takes interpretation of the
  interpretation of UTS #18.
  http://www.unicode.org/reports/tr18/

* Turning on/off indentation warnings with directives.
  ("# -*- warn-indent: true -*-" / "# -*- warn-indent: false -*-")

=== Compatibility issues (excluding feature bug fixes)

  * Rational#to_d

    See above.