= 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.

* REXML

  * REXML::Document.entity_expansion_limit=

    New method to set the entity expansion limit. By default the limit is
    set to 10000.  See the following URL for details.

    http://www.ruby-lang.org/en/news/2008/08/23/dos-vulnerability-in-rexml/

== Changes since the 1.8.6 release

=== Configuration changes

* vendor_ruby directory

  A new library directory named `vendor_ruby' is introduced in
  addition to `site_ruby'.  The idea is to separate libraries
  installed by the package system (`vendor') from manually (`site')
  installed libraries preventing the former from getting overwritten
  by the latter, while preserving the user option to override vendor
  libraries with site libraries. (`site_ruby' takes precedence over
  `vendor_ruby')

  If you are a package maintainer, make each library package configure
  the library passing the `--vendor' option to `extconf.rb' so that
  the library files will get installed under `vendor_ruby'.

  You can change the directory locations using configure options such
  as `--with-sitedir=DIR' and `--with-vendordir=DIR'.

=== Global constants

* new constants

  * RUBY_COPYRIGHT
  * RUBY_DESCRIPTION

=== Library updates (outstanding ones only)

* new library

  * securerandom

* builtin classes

  * Array#flatten
  * Array#flatten!

    Takes an optional argument that determines the level of recursion
    to flatten.

  * Array#eql?
  * Array#hash
  * Array#==
  * Array#<=>

    Handle recursive data properly.

  * Array#index
  * Array#rindex

    Take a block instead of an argument.

  * Array#collect!
  * Array#map!
  * Array#each
  * Array#each_index
  * Array#reverse_each
  * Array#reject
  * Array#reject!
  * Array#delete_if
  * Array#select

    Return an enumerator if no block is given.

    Note that #map and #collect still return an array unlike Ruby 1.9
    to keep compatibility.

  * Array#pop
  * Array#shift

    Take an optional argument specifying the number of elements to
    remove.

  * Array#choice
  * Array#combination
  * Array#cycle
  * Array#drop
  * Array#drop_while
  * Array#permutation
  * Array#product
  * Array#shuffle
  * Array#shuffle!
  * Array#take,
  * Array#take_while

    New methods.

  * Binding#eval

    New method.

  * Dir#each
  * Dir#foreach

    Return an enumerator if no block is given.

  * Enumerable::Enumerator

    New class for various enumeration defined by the enumerator library.

  * Enumerable#each_slice
  * Enumerable#each_cons
  * Object#to_enum
  * Object#enum_for

    New methods for various enumeration defined by the enumerator library.

  * Enumerable#count
  * Enumerable#cycle
  * Enumerable#drop
  * Enumerable#drop_while
  * Enumerable#find_index
  * Enumerable#first
  * Enumerable#group_by
  * Enumerable#max_by
  * Enumerable#min_by
  * Enumerable#minmax
  * Enumerable#minmax_by
  * Enumerable#none?
  * Enumerable#one?
  * Enumerable#take
  * Enumerable#take_while

    New methods.

  * Enumerable#find
  * Enumerable#find_all
  * Enumerable#partition
  * Enumerable#reject
  * Enumerable#select
  * Enumerable#sort_by

    Return an enumerator if no block is given.

    Note that #map and #collect still return an array unlike Ruby 1.9
    to keep compatibility.

  * Enumerable#inject

    Accepts a binary operator instead of a block.

  * Enumerable#reduce

    New alias to #inject.

  * Enumerable#to_a

    Can take optional arguments and pass them to #each.

  * Hash#eql?
  * Hash#hash
  * Hash#==

    Handle recursive data properly.

  * Hash#delete_if
  * Hash#each
  * Hash#each_key
  * Hash#each_pair
  * Hash#each_value
  * Hash#reject!
  * Hash#select
  * ENV.delete_if
  * ENV.each
  * ENV.each_key
  * ENV.each_pair
  * ENV.each_value
  * ENV.reject!
  * ENV.select

    Return an enumerator if no block is given.

  * GC.stress
  * GC.stress=

    New methods.

  * Integer#ord
  * Integer#odd?
  * Integer#even?
  * Integer#pred

    New methods.

  * Integer#downto
  * Integer#times
  * Integer#upto

    Return an enumerator if no block is given.

  * IO#each
  * IO#each_line
  * IO#each_byte
  * IO.foreach
  * ARGF.each
  * ARGF.each_line
  * ARGF.each_byte

    Return an enumerator if no block is given.

  * IO#bytes
  * IO#chars
  * IO#each_char
  * IO#getbyte
  * IO#lines
  * IO#readbyte
  * ARGF.bytes
  * ARGF.chars
  * ARGF.each_char
  * ARGF.getbyte
  * ARGF.lines
  * ARGF.readbyte

    New methods. 

  * Method#name
  * Method#owner
  * Method#receiver
  * UnboundMethod#name
  * UnboundMethod#owner

    New methods.

  * Module#class_exec
  * Module#module_exec

    New methods.

  * Numeric#step

    Return an enumerator if no block is given.

  * Object#instance_exec
  * Object#tap

    New methods.

  * ObjectSpace.each_object

    Return an enumerator if no block is given.

  * Process.exec implemented.

  * Range#each
  * Range#step

    Return an enumerator if no block is given.

  * Regexp.union accepts an array of patterns.

  * String#bytes

    New method

  * String#bytesize

    New method, returning the size in bytes. (alias length and size)

  * String#chars
  * String#each_char
  * String#lines
  * String#partition
  * String#rpartition
  * String#start_with?
  * String#end_with?

    New methods.  These are $KCODE aware unlike #index, #rindex and
    #include?.

  * String#each_byte
  * String#each
  * String#each_line
  * String#gsub(pattern)

    Return an enumerator if no block is given.

  * String#upto

    An optional second argument is added to specify if the last value
    should be included.

  * StopIteration

    New exception class that causes Kernel#loop to stop iteration when
    raised.

  * Struct#each
  * Struct#each_pair

    Return an enumerator if no block is given.

  * Symbol#to_proc

    New method.

  * __method__

    New global function that returns the name of the current method as
    a Symbol.

* enumerator

  * Enumerator is now a built-in module.  The #next and #rewind
    methods are implemented using the "generator" library.  Use with
    care and be aware of the performance loss.

* ipaddr

  * New methods
    * IPAddr#<=>
    * IPAddr#succ

      IPAddr objects are now comparable and enumerable having these
      methods.  This also means that it is possible to have a Range
      object between two IPAddr objects.

    * IPAddr#to_range

      A new method to create a Range object for the (network) address.

  * Type coercion support
    * IPAddr#&
    * IPAddr#|
    * IPAddr#==
    * IPAddr#include?

      These methods now accept a string or an integer instead of an
      IPAddr object as the argument.

* net/smtp

  * Support SSL/TLS.

* openssl

  * New classes
    * OpenSSL::PKey::EC
    * OpenSSL::PKey::EC::Group
    * OpenSSL::PKey::EC::Point
    * OpenSSL::PKey::PKCS5
    * OpenSSL::SSL::Session

  * Documentation!

  * Various new methods (see documentation).

  * Remove redundant module namespace in Cipher, Digest, PKCS7, PKCS12.
    Compatibility classes are provided which will be removed in Ruby 1.9.

* shellwords

  * Add methods for escaping shell-unsafe characters:
    * Shellwords.join
    * Shellwords.escape
    * Array#shelljoin
    * String#shellescape

  * Add shorthand methods:
    * Shellwords.split (alias shellwords)
    * String#shellsplit

* stringio

  * StringIO#getbyte
  * StringIO#readbyte

    New methods. (aliases for compatibility with 1.9)

  * StringIO#each_char
  * StringIO#chars

    New methods.

  * StringIO#each
  * StringIO#each_line
  * StringIO#each_byte

    Return an enumerator if no block is given.

* tempfile

  * Tempfile.open and Tempfile.new now accept a suffix for the
    temporary file to be created.  To specify a suffix, pass an array
    of [basename, suffix] as the first argument.

      Tempfile.open(['image', 'jpg']) { |tempfile| ... }

* tmpdir

  * New method:

    * Dir.mktmpdir

* uri

  * added LDAPS scheme.
  * Change for RFC3986:
    * FTP
      * URI('ftp://example.com/foo').path #=> 'foo'
      * URI('ftp://example.com/%2Ffoo').path #=> '/foo'
      * URI::FTP.build([nil, 'example.com', nil, '/foo', 'i').to_s #=> 'ftp://example.com/%2Ffoo;type=i'
    * URI merge
      * URI('http://a/b/c/d;p?q').merge('?y') == URI('http://a/b/c/d;p?y')
      * URI('http://a/b/c/d;p?q').merge('/./g') == URI('http://a/g')
      * URI('http://a/b/c/d;p?q').merge('/../g') == URI('http://a/g')
      * URI('http://a/b/c/d;p?q').merge('../../../g') == URI('http://a/g')
      * URI('http://a/b/c/d;p?q').merge('../../../../g') == URI('http://a/g')

* rss

  * 0.1.6 -> 0.2.4

  * Fix image module URI

  * Atom support

  * ITunes module support

  * Slash module support

  * content:encoded with RSS 2.0 support

=== Interpreter Implementation

* passing a block to a Proc [experimental]

  This implementation in current shape is known to be buggy/broken,
  especially with nested block invocation.  Take this as an
  experimental feature.

* stack trace

  On non-SystemStackError exception, full stack trace is shown.

=== Compatibility issues (excluding feature bug fixes)

* String#slice! had some unintentional bugs and they have been fixed
  because either they disagreed with documentation or their respective
  behavior of #slice.  Unfortunately, this causes some
  incompatibilities in the following (somewhat rare) cases.

  * #slice! no longer expands the array when an out-of-boundary value
    is given.

      # Ruby 1.8.6
      a = [1,2]
      a.slice!(4,0)   #=> nil
      a               #=> [1,2,nil,nil]

      # Ruby 1.8.7
      a = [1,2]
      a.slice!(4,0)   #=> nil
      a               #=> [1,2]

  * #slice! no longer raises an exception but returns nil when a
    negative length or out-of-boundary negative position is given.

      # Ruby 1.8.6
      a = [1,2]
      a.slice!(1,-1)  #=> (raises IndexError)
      a.slice!(-5,1)  #=> (raises IndexError)

      # Ruby 1.8.7
      a = [1,2]
      a.slice!(1,-1)  #=> nil
      a.slice!(-5,1)  #=> nil

* String#to_i, String#hex and String#oct no longer accept a sequence
  of underscores (`__') as part of a number.

    # Ruby 1.8.6
    '1__0'.to_i     #=> 10
    '1__0'.to_i(2)  #=> 2  # 0b10
    '1__0'.oct      #=> 8  # 010
    '1__0'.hex      #=> 16 # 0x10

    # Ruby 1.8.7
    '1__0'.to_i     #=> 1
    '1__0'.to_i(2)  #=> 1
    '1__0'.oct      #=> 1
    '1__0'.hex      #=> 1

  The old behavior was inconsistent with Ruby syntax and considered as
  a bug.

* date

  * Date.parse

    '##.##.##' (where each '#' is a digit) is now taken as 'YY.MM.DD'
    instead of 'MM.DD.YY'.  While the change may confuse you, you can
    always use Date.strptime() when you know what you are dealing
    with.

* stringio

  * StringIO#each_byte

    The return value changed from nil to self.  This is what the
    document says and the same as each_line() does.

* tempfile

  * The file name format has changed.  No dots are included by default
    in temporary file names any more.  See above for how to specify a
    suffix.

* uri

  * See above for details.

== Changes since the 1.8.5 release

=== New platforms/build tools support

* IA64 HP-UX

* Visual C++ 8 SP1

* autoconf 2.6x

=== Global constants

* RUBY_PATCHLEVEL

  New constant since 1.8.5-p1.

=== Library updates (outstanding ones only)

* builtin classes

  * New method: Kernel#instance_variable_defined?

  * New method: Module#class_variable_defined?

  * New feature: Dir::glob() can now take an array of glob patterns.

* date

  * Updated based on date2 4.0.3.

* digest

  * New internal APIs for C and Ruby.

  * Support for autoloading.

      require 'digest'

      # autoloads digest/md5
      md = Digest::MD5.digest("string")

  * New digest class methods: file

  * New digest instance methods: clone, reset, new, inspect,
    digest_length (alias size or length), block_length()

  * New library: digest/bubblebabble

  * New function: Digest(name)

* fileutils

  * New option for FileUtils.cp_r(): :remove_destination

* nkf

  * Updated based on nkf as of 2007-01-28.

* thread

  * Replaced with much faster mutex implementation in C.  The former
    implementation, which is slow but considered to be stable, is
    available with a configure option `--disable-fastthread'.

* tk

  * Updated Tile extension support based on Tile 0.7.8.

  * Support --without-X11 configure option for non-X11 versions of
    Tcl/Tk (e.g. Tcl/Tk Aqua).

  * New sample script: irbtkw.rbw -- IRB on Ruby/Tk. It has no trouble
    about STDIN blocking on Windows.

* webrick

  * New method: WEBrick::Cookie.parse_set_cookies()

=== Compatibility issues (excluding feature bug fixes)

* builtin classes

  * String#intern now raises SecurityError when $SAFE level is greater
    than zero.

* date

  * Time#to_date and Time#to_datetime are added as private methods.
    They cause name conflict error in ActiveSupport 1.4.1 and prior,
    which comes with Rails 1.2.2 and prior.  Updating ActiveSupport
    and/or Rails to the latest versions fixes the problem.

* digest

  * The constructor does no longer take an initial string to feed.
    The following examples show how to migrate:

      # Before
      md = Digest::MD5.new("string")
      # After (works with any version)
      md = Digest::MD5.new.update("string")
        
      # Before
      hd = Digest::MD5.new("string").hexdigest
      # After (works with any version)
      hd = Digest::MD5.hexdigest("string")

* fileutils

  * A minor implementation change breaks Rake <=0.7.1.
    Updating Rake to 0.7.2 or higher fixes the problem.

* tk

  * Tk::X_Scrollable (Y_Scrollable) is renamed to Tk::XScrollable
    (YScrollable). Tk::X_Scrollable (Y_Scrollable) is still available,
    but it is an alias name.