Update with branches/ruby_1_9_1/NEWS.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@22267 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This commit is contained in:
knu 2009-02-12 15:04:41 +00:00
Родитель 643b8cb4b4
Коммит 385b6a15ff
1 изменённых файлов: 298 добавлений и 82 удалений

Просмотреть файл

@ -8,6 +8,7 @@ reference information is supplied with. For a full list of changes
with all sufficient information, see the ChangeLog file.
== Changes since the 1.8.7 release
See doc/NEWS-1.8.7 for changes between 1.8.6 and 1.8.7.
=== Compatibility issues
@ -15,144 +16,273 @@ with all sufficient information, see the ChangeLog file.
* New syntax and semantics
o Block arguments are always local
o Block arguments
o New semantics for block arguments
o Block local variables
o Method used for splat arguments: #to_splat
o defined? and local variables
o Parser expects that your source code has only valid byte
sequence in some character encoding. Use magic comments
to tell the parser which encoding you use.
o New semantics for constant definition in instance_eval
or in module_eval.
* Deprecated syntax
o colon (:) instead of "then" in if/unless or case expression.
o retry in a loop or an iterator.
* builtin classes and objects
* Kernel and Object
o Kernel#require
o Kernel#methods and #singleton_methods used to return an
array of strings but now they return an array of symbols.
* Class and Module
o Class variables behavior
o Module#attr is an alias of attr_reader
o Module#instance_methods, #private_instance_methods, #public_instance_methods
o Module#attr works as Module#attr_reader by default.
Optional boolean argument is obsolete.
o Module#instance_methods, #private_instance_methods and
#public_instance_methods used to return an array of
strings but now they return an array of symbols.
o Extra subclassing check when binding UnboundMethods
* Exceptions
o Equality of exceptions
o SystemStackError
o SecurityError
o Exceptions are equal to each other if they belong to
the same class and have the same message and backtrace.
o SystemStackError used to be a subclass of StandardError
but not it is a direct subclass of Exception.
o SecurityError: ditto
o Removed Exception#to_str [Ruby2]
* Enumerable and Enumerator
o Enumerable::Enumerator, compatibility alias of Enumerator,
is removed.
o Enumerable#{map,collect} called without a block returns
an enumerator.
o Even more builtin and bundled libraries have been made to
return an enumerator when called without a block.
* Array
o Array#nitems was removed (use count {|i| !i.nil?})
o Array#choice was removed (use sample)
o Array#[m,n] = nil places nil in the array.
* Enumerable and Enumerator
o Enumerable#map,collect called without a block returns
an enumerator.
* File and Dir operations
o #to_path in File.path, File.chmod, File.lchmod, File.chown, File.lchown, File.utime, File.unlink... [Ruby2]
o Dir.[], Dir.glob
o Dir.exist?
* String
o No longer an Enumerable
o ?c semantics
o "One-char-wide" semantics for String#[] and String#[]= [Ruby2]
o Encoding-awareness
* Hash
o Hash#to_s is equivalent to Hash#inspect
o Semantics for Hash#each and Hash#each_pair
o Hash#select returns a hash
o Hash#key is the new name for #index which has been deprecated.
o Most of the changes in Hash apply to hash like interfaces such as
ENV and *DBM.
* Integer
o Integer(nil) raises TypeError
o Hash#key is the new name for #index which has been
deprecated.
o Hash preserves order. It enumerates its elements in the
order in which the keys are inserted.
o Most of the changes in Hash apply to hash like interfaces
such as ENV and *DBM.
* IO operations
o Many methods used to act byte-wise but now some of those act
character-wise. You can use alternate byte-wise methods.
o IO#getc
o Non-blocking IO
o Kernel#open takes "t" for newline conversion
o Kernel#open takes encoding specified
o IO#initialize now accepts an IO argument
o IO automatically converts byte sequence from a character
encodings into another if specified.
o StringIO#readpartial
o IO.try_convert
o Limit input in IO#gets, IO#readline, IO#readlines, IO#each_line, IO#lines, IO.foreach, IO.readlines, StringIO#gets, StringIO#readline, StringIO#each, StringIO#readlines
o IO.binread
o IO.copy_stream
o IO#binmode?
o IO#close_on_exec= and IO#close_on_exec?
o Limit input in IO#gets, IO#readline, IO#readlines,
IO#each_line, IO#lines, IO.foreach, IO.readlines,
StringIO#gets, StringIO#readline, StringIO#each,
StringIO#readlines
o IO#ungetc, StringIO#ungetc
o IO#internal_encoding, IO#external_encoding, IO#set_encoding
o IO.pipe takes encoding option
o IO#ungetbyte, StringIO#ungetbyte
o IO#internal_encoding, IO#external_encoding,
IO#set_encoding
o IO.pipe takes encoding option
o Directive %u behaves like %d for negative values in
printf-style formatting.
* File and Dir operations
o #to_path is called as necessary in File.path, File.chmod,
File.lchmod, File.chown, File.lchown, File.utime,
File.unlink, etc..
o File.world_readable?
o File.world_writable?
o Dir.[], Dir.glob
o Dir.exist?
o Dir.exists?
* File::Stat
o File::Stat#world_readable?
o File::Stat#world_writable?
* String
o No longer an Enumerable: use each_line/lines for line
oriented operation
o Encoding-awareness
o Character-wise semantics in many methods instead of
byte-wise.
o String#[]: Indexing a String with an integer returns a
single character String instead of an integer.
o String#[]=: No longer takes an integer as right
side value. Note that "str[i] = ?c" because of
the following change.
o ?c is evaluated to a single character string
instead of an integer.
* Regexp
o Encoding-awareness
o Regexp matches only with strings which is encoded in a
compatible character encoding to the regexp's.
o Regexp#kcode is removed. use Regexp#encoding.
* Symbols: restriction on literal symbols
* Numeric
o Numeric#div always rounds as Integer#div has done.
o Numeric#fdiv: ditto.
* Integer
o Integer(nil) raises TypeError
* Fixnum
o Fixnum#id2name removed
o Fixnum#to_sym removed
* Struct
o Struct#inspect
* Time
o New format in Time#to_s
o Timezone information preserved on Marshal.dump/load
* Struct
o Struct#inspect
* Symbols: restriction on literal symbols
* $SAFE and bound methods
o New trusted/untrusted model in addition to
tainted/untainted model.
* Deprecation
o VERSION and friends
o Kernel.to_a
o Kernel#getc
o $= (global flag for case-sensitiveness on string matching)
o Kernel#to_a
o Kernel#getc, #gsub, #sub
o Kernel#callcc and Continuation now become 'continuation'
bundled library.
o Object#type
o Removed Array and Hash #indices, #indexes
o Hash#index
o ENV.index
o Symbol#to_int
o Removed Array and Hash #indices, #indexes
o libraries: base64, mailread, getopts, parsearg, soap, wsdl
o Process::Status#to_int
o Numeric#rdiv
o Precision is removed. Don't cry, it will be redesigned
and come back in future version.
o Symbol#to_int and Symbol#to_i
o $KCODE is no longer effective. Use Encoding related
features of each class.
o VERSION and friends
* bundled libraries
* Pathname
o No longer has #to_str nor #=~.
* time and date
o Time.parse and Date.parse interprets slashed numerical dates
as "dd/mm/yyyy".
* Readline
o If Readline uses libedit, Readline::HISTORY[0] returns the first of the history.
o If Readline uses libedit, Readline::HISTORY[0] returns the
first of the history.
* Continuation
o as above
* Deprecation
o Complex#image: use Complex#imag
o All SSL-related class methods in Net::SMTP
o Prime#cache, Prime#primes, Prime#primes_so_far
o mailread library: use tmail gem.
o cgi-lib library: use cgi.
o date2 library: use date.
o eregex library
o finalize library: use ObjectSpace.define_finalizer if you
really need a finalizer. really?
o ftools library: use fileutils.
o generator library: use Enumerator.
o importenv library and Env library
o jcode library: use multilingualization support of String
o parsedate library
o ping library
o readbytes library
o getopts library and parsearg library: use optparse or
getoptlong.
o soap, wsdl and xsd libraries: use soap4r gem.
o Win32API library: use dl.
o dl library: Reimplemented and API changed. use the new
version of dl or ffi gem.
o rubyunit library and runit library: use minitest or
test/unit. Or use anything you love through RubyGems.
o test/unit is reimplemented on top of minitest. This is
not fully compatible with the original.
=== Language core changes
* New syntax and semantics
o New literal hash syntax
o Magic comments to declare in which encoding your source
code is written
o New literal hash syntax and new syntax for hash style
arguments
o New syntax for lambdas
o .() and calling Procs without #call/#[]
o Block in block arguments
o Block local variables
o Mandatory arguments after optional arguments allowed
o Multiple splats allowed
o Arguments to #[]
o printf-style formatted strings (%)
o Newlines allowed before ternary colon
o #[] can take splatted arguments, hash style arguments
and a block.
o New directives in printf-style formatted strings (%).
o Newlines allowed before ternary colon operator (:) and
method call dot operator (.)
o Negative operators such as !, != and !~ are now
overloadable
o Encoding.default_external and default_internal
o __ENCODING__: New pseudo variable to hold the current
script's encoding
=== Library updates
* builtin classes and objects
* Kernel and Object
o BasicObject
o Object#=~
o Kernel#instance_variable_defined?
o Object#=~ returns nil instead of false by default.
o Kernel#define_singleton_method
o Kernel#singleton_methods, Kernel#methods
o Kernel#load can load a library from the highest versions
of gems by default.
* Class and Module
o Module#const_defined?, #const_get and #method_defined?
o Module#class_variable_defined?
o #class_variable_{set,get}
take an optional parameter.
o #class_variable_{set,get} are public.
o Class of singleton classes
* Errno::EXXX
o All of those are always defined. Errno::EXXX will be
defined as an alias to Errno::NOERROR if your platform
does not have one.
* Binding#eval
* Blocks and Procs
o Arity of blocks without arguments
o proc is now a synonym of Proc.new
* Blocks and Procs
o Proc#yield
o Passing blocks to #[]
o Proc#lambda?
o Proc#curry
* Fiber: coroutines/micro-threads
* Thread
o Thread.critical and Thread.critical= removed
o Thread#exit!, Thread#kill! and Thread#terminate! removed.
* Enumerable and Enumerator
o Enumerable#each_with_object [experimental]
o Enumerator#enum_cons and Enumerator#enum_slice are
removed. Use #each_cons and #each_slice without a block.
o Enumerable#each_with_index can take optional arguments
and passes them to #each.
o Enumerable#each_with_object
o Enumerator#with_object
o Enumerator.new { ... }
* Regexp#match, String#match
o Regexp#match, String#match
* Encoding
* Encoding::Converter
* Fiber: coroutines/micro-threads
* Array
o Array#delete returns a deleted element rather than a given
object
o Array#to_s is equivalent to Array#inspect
o Array.try_convert
o Array#pack('m0') complies with RFC 4648.
* Hash
o preserving item insertion order
o Hash#default_proc=
o Hash#_compare_by_identity and Hash#compare_by_identity?
o Hash.try_convert
* Numeric
o Numeric#upto, #downto, #times, #step
o Numeric#real?, Complex#real?
o Numeric#div
o Numeric#fdiv
o Hash#assoc
o Hash#rassoc
o Hash#flatten
* Range
o Range#cover?
o Range#include? iterates over elements and compares the
@ -160,45 +290,131 @@ with all sufficient information, see the ChangeLog file.
Use Range#cover? for the old behavior, i.e. comparison
with boundary values.
o Range#min, Range#max
* Regexp
o Regexp#=== matches symbols
o Regexp.try_convert
* File and Dir operations
o New methods
* Process
o Process.spawn
o Process.daemon
* String
o String#clear
o String#ord
o String#getbyte, String#setbyte
o String#encoding
o String#force_encoding
o String#chars and String#each_char act as character-wise.
o String#codepoints, String#each_codepoint
o String#unpack with a block
o String#hash
o String.try_convert
o String.encoding, String#force_encoding, String#encode
o String#encoding
o String#force_encoding, String#encode and String#encode!
o String#ascii_only?
o String#valid_encoding?
o String#match
* Symbol
o Zero-length symbols allowed
o Symbol#=== matches strings
o Symbol#intern
o Symbol#encoding
o Symbol methods similar to those in String
* Regexp
o Regexp#=== matches symbols
o Regexp.try_convert
o Regexp#match
o Regexp#fixed_encoding?
o Regexp#encoding
o Regexp#named_captures
o Regexp#names
* MatchData
o MatchData#names
o MatchData#regexp
* Encoding
* Encoding::Converter
o supports conversion between many encodings
* Numeric
o Numeric#upto, #downto, #times, #step
o Numeric#real?, Complex#real?
o Numeric#magnitude
* Rational / Complex
o They are in the core library now
* Math
o Math#log and Math#log2
* File and Dir operations
o New methods
* Process
o Process.daemon
* Readline
o Readline.vi_editing_mode?
o Readline.emacs_editing_mode?
o Readline::HISTORY.clear
o Math#log takes an optional argument.
o Math#log2
o Math#cbrt, Math#lgamma, Math#gamma
* Time
o Time.times removed. Use Process.times.
o Time#sunday?
o Time#monday?
o Time#tuesday?
o Time#wednesday?
o Time#thursday?
o Time#friday?
o Time#saturday?
o Time#tv_nsec and Time#nsec
* Misc. new methods
o RUBY_ENGINE to distinguish between Ruby processor implementation
o public_method
o public_send
o GC.count
o ObjectSpace.count_objects
o Method#hash, Proc#hash
o Method#source_location, UnboundMethod#source_location and
Proc#source_location
o __callee__
o Elements in $LOAD_PATH and $LOADED_FEATURES are expanded
* bundled libraries
* RubyGems
o Package management system for Ruby.
o Integrated with Ruby's library loader.
* Rake
o Ruby make. A simple ruby build program with capabilities
similar to make.
* minitest
o Our new testing library which is faster, cleaner and easier
to read than the old test/unit.
o You can introduce the old test/unit as testunit gem through
RubyGems if you want.
* CMath
o Complex number version of Math
* Prime
o Extracted from Mathn and improved. You can easily enumerate
prime numbers.
o Prime.new is obsolete. Use its class methods.
* ripper
o Ruby script parser
* Readline
o Readline.vi_editing_mode?
o Readline.emacs_editing_mode?
o Readline::HISTORY.clear
* Tk
o TkXXX widget classes are removed and redefined as aliases of
Tk::XXX classes.
* RDoc
o Updated to version 2.2.2. See:
http://rubyforge.org/frs/shownotes.php?group_id=627&release_id=26434
* commandline options
* -E, --encoding
* -U
* --enable-gems, --disable-gems
* --enable-rubyopt, --disable-rubyopt
* long options are allowed in RUBYOPT environment variable.
=== Implementation changes
* Memory Diet
* Object Compaction - Object, Array, String, Hash, Struct, Class, Module
* st_table compaction (inlining small tables)
* YARV
* Memory Diet
* Object Compaction - Object, Array, String, Hash, Struct, Class,
Module
* st_table compaction (inlining small tables)
* YARV
* Ruby codes are compiled into opcodes before executed.
* Native thread
* Platform supports
* Support levels
(0) Supported
(1) Best effort
(2) Perhaps
(3) Not supported
* Dropped
o No longer supports djgpp, bcc32, human68k, MacOS 9 or earlier,
VMS nor Windows CE.