planned versions

• one major release a year

• a few minor releases every 2-3 months

short depreciation cycle

• warning during at least one major cycle

• effective removal at the next major release

• modules removed from the core go on the CPAN

5.12.0 on 2010.04.12

5.12.1 on 2010.05.16

5.12.2 on 2010.09.06

5.12.3 on 2011.01.21

5.12.4 on 2011.06.20

5.14.0 on 2011.05.14

5.14.1 on 2011.06.17

5.14.2 on 2011.09.26

5.16.0 on 2012.05.12

• 5.12.x is no longer supported

5.16.1 on 2012.08.08

5.16.2 on 2012.11.01

5.18.0 on 2012.05

current_sub

• __SUB__ keyword

• reference to the current sub-routine

• works in special blocks (5.16.1)

more keywords are now available in CORE::

unicode_eval

• eval() always evaluates the given string as Unicode

evalbytes

• evalbytes() always evaluates the given string as bytes

substr() can now return a lvalue

 my $string = "The quick brown fox jumps over the lazy dog";
 my $mask1  = \substr($string, 4, 5);
 my $mask2  = \substr($string, -8);
 print "$$mask1 | $$mask2\n";  # prints "quick | lazy dog"

 $string = "Le vif zéphyr souffle sur ces kumquats disgracieux";
 print "$$mask1 | $$mask2\n";  # prints "if zé | gracieux"

tied now returns the actual scalar that holds the object to which the variable is tied

 Scalar::Util::weaken(tied $tied_variable);

Unicode 6.1.0

except U+1F514 (BELL), which conflicts with U+0007 (BELL, ALERT, BEL) ; will be renamed in 5.18

some names and properties have been renamed

Perl accepts both old and new names

charnames autoloaded on the first \N{...}

\N{...} now supports loose match

full Unicode support for all symbol names (methods, typeglobs, packages...)

full case folding

gives better results when comparing two strings case-insensitively

now available via the fc() function and the \F sequence

 lc($str1) eq lc($str2)  # incorrect
 uc($str1) eq uc($str2)  # incorrect

 $str1 =~ /\Q$str2/i     # correct, but uses a regexp

 fc($str1) eq fc($str2)  # correct

arybase - implementation of $[

the values of $$, $<, $>, $(, $) are no longer cached

$$ can now be assigned to

$^X is now converted to an absolute path on OS X, FreeBSD and Solaris 10 & 11

the \$ and \[$] prototypes now accept any scalar lvalue argument

the _ prototype is now allowed before @ or %

empty DESTROY methods are no longer called (5.12)

DESTROY methods beginning with a return are no longer called

assignment to substr() in void context is now twice its previous speed

faster use VERSION, no longer loads feature.pm

faster glob() (but not on VMS)

faster search on Unicode properties in regexps

chained short-circuit operators (&&, ||, //) are now faster due to reduced optree traversal

Devel::DProf, Shell

Perl4 libraries (assert.pl, bigint.pl, getcwd.pl, getopt.pl hostname.pl, open2.pl, tainted.pl, etc)

swapping of $< and $>

sfio support

(as compiled from the perldelta of 5.7.0 ~ 5.7.8)

even more keywords in CORE::

kill now supports negative signal names: KILL -HUP -$$

next, last and redo now supports computed labels

given now aliases the global $_, like foreach

${^LAST_FH} contains the filehandle that was last read

lexical subroutines (experimental)

 use feature "lexical_subs";

 state sub myfunc { ... }
 my sub myfunc { ... }
 our sub myfunc { ... }

warnings on experimental features

 no warnings "experimental::lexical_subs";
 use feature "lexical_subs";

qw(..) no longer fools the parser

•  for my $pony qw< Twilight Fluttershy Pinkie_Pie > { ... }  # WRONG

•  for my $pony (qw< Twilight Fluttershy Pinkie_Pie >) { ... }

stricter parsing of s///e

$&, $` and $' are no longer slow

\s now matches a vertical tab (\v)

literal unescaped { is now deprecated

• must be escaped or part of a construct

• catches typos like /\w{,5}/

• provisioned for future syntax extensions

/(?{})/ and /(??{})/ have been heavily reworked

regular expression set operations (?[ ])

experimental: no warnings "experimental::regex_sets";

combines classes with & (intersection), + or | (union), - (substraction), ^ (symmetric difference)

 # match alphabetic characters, no digits
 /(?[ \w - \d ])/

 # match all the digits in the Thai or Laotian scripts
 /(?[ ( \p{Thai} + \p{Lao} ) & \p{Digit} ])/

readline() with $/ = \N now reads N characters, not N bytes

Unicode 6.2

\N{BELL} now refers to U+1F514 instead of U+0007

study now is a complete no-op

require now dies for unreadable files

do now always sets $@ (like eval)

do function now uses as a file name what function returns

several fixes on the formats

new hash function: Murmurhash-32 v3

new copy-on-write mechanism

in regexps, matching against Unicode properties is faster

pack with constant arguments is now constant folded when possible

.. x .. is now constant folded at compile time when possible

syntax

• lexical $_ (too confusing, too complex)

modules

• CPANPLUS and its deps, Archive::Extract, Locale::Maketext::Simple, Log::Message, Log::Message::Simple, Module::Pluggable, Object::Accessor, Params::Check, Term::UI

• Filter::Simple, PerlIO::mmap

• Pod::LaTeX, Pod::Parser, SelfLoader

• B::Lint, Text::Soundex, Thread.pm

platforms

• EPOC, MPE/iX, UTS Global, VM/ESA, Rhapsody

platforms

• BeOS, djgpp, dgux

compilers

• Borland