Fetching contributors…
Cannot retrieve contributors at this time
888 lines (614 sloc) 27.6 KB
perl5i - Fix as much of Perl 5 as possible in one pragma
use perl5i::2;
$ perl5i
Perl 5 has a lot of warts. There's a lot of individual modules and
techniques out there to fix those warts. perl5i aims to pull the best of
them together into one module so you can turn them on all at once.
This includes adding features, changing existing core functions and
changing defaults. It will likely not be 100% backwards compatible with
Perl 5, though it will be 99%, perl5i will try to have a lexical effect.
Please add to this imaginary world and help make it real, either by
telling me what Perl looks like in your imagination
( or make a fork (forking on
github is like a branch you control) and implement it yourself.
Using perl5i
Because perl5i *plans* to be incompatible in the future, you do not
simply `use perl5i'. You must declare which major version of perl5i you
are using. You do this like so:
# Use perl5i major version 2
use perl5i::2;
Thus the code you write with, for example, `perl5i::2' will always
remain compatible even as perl5i moves on.
If you want to be daring, you can `use perl5i::latest' to get the latest
If you want your module to depend on perl5i, you should depend on the
versioned class. For example, depend on `perl5i::2' and not `perl5i'.
See VERSIONING for more information about perl5i's versioning scheme.
What it does
perl5i enables each of these modules and adds/changes these functions.
We'll provide a brief description here, but you should look at each of
their documentation for full details.
The Meta Object
Every object (and everything is an object) now has a meta object
associated with it. Using the meta object you can ask things about the
object which were previously over complicated. For example...
# the object's class
my $class = $obj->mo->class;
# its parent classes
my @isa = $obj->mo->isa;
# the complete inheritance hierarchy
my @complete_isa = $obj->mo->linear_isa;
# the reference type of the object
my $reftype = $obj->mo->reftype;
A meta object is used to avoid polluting the global method space. `mo'
was chosen to avoid clashing with Moose's meta object.
See perl5i::Meta for complete details.
Subroutine and Method Signatures
perl5i makes it easier to declare what parameters a subroutine takes.
func hello($place) {
say "Hello, $place!\n";
method get($key) {
return $self->{$key};
method new($class: %args) {
return bless \%args, $class;
`func' and `method' define subroutines as `sub' does, with some extra
The signature syntax is currently very simple. The content will be
assigned from @_. This:
func add($this, $that) {
return $this + $that;
is equivalent to:
sub add {
my($this, $that) = @_;
return $this + $that;
`method' defines a method. This is the same as a subroutine, but the
first argument, the *invocant*, will be removed and made into `$self'.
method get($key) {
return $self->{$key};
sub get {
my $self = shift;
my($key) = @_;
return $self->{$key};
Methods have a special bit of syntax. If the first item in the siganture
is `$var:' it will change the variable used to store the invocant.
method new($class: %args) {
return bless $class, \%args;
is equivalent to:
sub new {
my $class = shift;
my %args = @_;
return bless $class, \%args;
Anonymous functions and methods work, too.
my $code = func($message) { say $message };
Guarantees include:
@_ will not be modified except by removing the invocant
Future versions of perl5i will add to the signature syntax and
capabilities. Planned expansions include:
Signature validation
Signature documentation
Named parameters
Required parameters
Read only parameters
Aliased parameters
Anonymous method and function declaration
Variable method and function names
Parameter traits
Traditional prototypes
See for
more details about future expansions.
The equivalencies above should only be taken for illustrative purposes,
they are not guaranteed to be literally equivalent.
Note that while all parameters are optional by default, the number of
parameters will eventually be enforced. For example, right now this will
func add($this, $that) { return $this + $that }
say add(1,2,3); # says 3
The extra argument is ignored. In future versions of perl5i this will be
a runtime error.
Signature Introspection
The signature of a subroutine defined with `func' or `method' can be
queried by calling the `signature' method on the code reference.
func hello($greeting, $place) { say "$greeting, $place" }
my $code = \&hello;
say $code->signature->num_positional_params; # prints 2
Functions defined with `sub' will not have a signature.
See perl5i::Signature for more details.
autobox allows methods to be defined for and called on most unblessed
variables. This means you can call methods on ordinary strings, lists
and hashes! It also means perl5i can add a lot of functionality without
polluting the global namespace.
autobox::Core wraps a lot of Perl's built in functions so they can be
called as methods on unblessed variables. `@a->pop' for example.
$scalar_reference->alias( @identifiers );
@alias->alias( @identifiers );
%hash->alias( @identifiers );
(\&code)->alias( @identifiers );
Aliases a variable to a new global name.
my $code = sub { 42 };
$code->alias( "foo" );
say foo(); # prints 42
It will work on everything except scalar references.
our %stuff;
%other_hash->alias( "stuff" ); # %stuff now aliased to %other_hash
It is not a copy, changes to one will change the other.
my %things = (foo => 23);
our %stuff;
%things->alias( "stuff" ); # alias %things to %stuff
$stuff{foo} = 42; # change %stuff
say $things{foo}; # and it will show up in %things
Multiple @identifiers will be joined with '::' and used as the fully
qualified name for the alias.
my $class = "Some::Class";
my $name = "foo";
sub { 99 }->alias( $class, $name );
say Some::Class->foo; # prints 99
If there is just one @identifier and it has no "::" in it, the current
caller will be prepended. `$thing->alias("name")' is shorthand for
`$thing->alias(CLASS, "name")'
Due to limitations in autobox, non-reference scalars cannot be aliased.
Alias a scalar ref instead.
my $thing = 23;
$thing->alias("foo"); # error
my $thing = \23;
$thing->alias("foo"); # $foo is now aliased to $thing
This is basically a nicer way to say:
no strict 'refs';
*{$package . '::'. $name} = $reference;
Scalar Autoboxing
perl5i adds some methods to scalars of its own.
my $centered_string = $string->center($length);
my $centered_string = $string->center($length, $character);
Centers $string between $character. $centered_string will be of length
`$character' defaults to " ".
say "Hello"->center(10); # " Hello ";
say "Hello"->center(10, '-'); # "---Hello--";
`center()' will never truncate `$string'. If $length is less than
`$string->length' it will just return `$string'.
say "Hello"->center(4); # "Hello";
my $rounded_number = $number->round;
Round to the nearest integer.
my $new_number = $number->round_up;
Rounds the $number towards infinity.
2.45->round_up; # 3
(-2.45)->round_up; # -2
ceil() is a synonym for round_up().
my $new_number = $number->round_down;
Rounds the $number towards negative infinity.
2.45->round_down; # 2
(-2.45)->round_down; # -3
floor() is a synonyn for round_down().
$is_a_number = $thing->is_number;
Returns true if $thing is a number understood by Perl.
12.34->is_number; # true
"12.34"->is_number; # also true
"eleven"->is_number; # false
$is_positive = $thing->is_positive;
Returns true if $thing is a positive number.
0 is not positive.
$is_negative = $thing->is_negative;
Returns true if $thing is a negative number.
0 is not negative.
$is_even = $thing->is_even;
Returns true if $thing is an even integer.
$is_odd = $thing->is_odd;
Returns true if $thing is an odd integer.
$is_an_integer = $thing->is_integer;
Returns true if $thing is an integer.
12->is_integer; # true
12.34->is_integer; # false
"eleven"->is_integer; # false
A synonym for is_integer
$is_a_decimal_number = $thing->is_decimal;
Returns true if $thing is a decimal number.
12->is_decimal; # false
12.34->is_decimal; # true
".34"->is_decimal; # true
"point five"->is_decimal; # false
my $module = $module->require;
Will `require' the given $module. This avoids funny things like `eval
qq[require $module] or die $@'. It accepts only module names.
On failure it will throw an exception, just like `require'. On a success
it returns the $module. This is mostly useful so that you can
immediately call $module's `import' method to emulate a `use'.
# like "use $module qw(foo bar);" if that worked
$module->require->import(qw(foo bar));
# like "use $module;" if that worked
my $wrapped = $string->wrap( width => $cols, separator => $sep );
Wraps $string to width $cols, breaking lines at word boundries using
separator $sep.
If no width is given, $cols defaults to 76. Default line separator is
the newline character "\n".
See Text::Wrap for details.
my $trimmed = $string->trim;
my $trimmed = $string->trim($character_set);
Trim whitespace. ltrim() trims off the start of the string (left),
rtrim() off the end (right) and trim() off both the start and end.
my $string = ' testme'->ltrim; # 'testme'
my $string = 'testme '->rtrim; # 'testme'
my $string = ' testme '->trim; # 'testme'
They all take an optional $character_set which will determine what
characters should be trimmed. It follows regex character set syntax so
`A-Z' will trim everything from A to Z. Defaults to `\s', whitespace.
my $string = '-> test <-'->trim('-><'); # ' test '
my $name = 'joe smith'->title_case; # Joe Smith
Will uppercase every word character that follows a wordbreak character.
my $module = $path->path2module;
Given a relative $path it will return the Perl module this represents.
For example,
"Foo/"->path2module; # "Foo::Bar"
It will throw an exception if given something which could not be a path
to a Perl module.
my $path = $module->module2path;
Will return the relative $path in which the Perl $module can be found.
For example,
"Foo::Bar"->module2path; # "Foo/"
my $number_grouped = $number->group_digits;
my $number_grouped = $number->group_digits(\%options);
Turns a number like 1234567 into a string like 1,234,567 known as "digit
It honors your current locale to determine the separator and grouping.
This can be overriden using `%options'.
NOTE: many systems do not have their numeric locales set properly
The character used to separate groups. Defaults to "thousands_sep"
in your locale or "," if your locale doesn't specify.
The decimal point character. Defaults to "decimal_point" in your
locale or "." if your locale does not specify.
How many numbers in a group? Defaults to "grouping" in your locale
or 3 if your locale doesn't specify.
Note: we don't honor the full grouping locale, its a wee bit too
If true, it will treat the number as currency and use the monetary
locale settings. "mon_thousands_sep" instead of "thousands_sep" and
"mon_grouping" instead of "grouping".
1234->group_digits; # 1,234 (assuming US locale)
1234->group_digits( separator => "." ); # 1.234
my $number_grouped = $number->commify;
my $number_grouped = $number->commify(\%options);
commify() is just like group_digits() but it is not locale aware. It is
useful when you want a predictable result regardless of the user's
locale settings.
`%options' defaults to `( separator => ",", grouping => 3, decimal_point
=> "." )'. Each key will be overriden individually.
1234->commify; # 1,234
1234->commify({ separator => "." }); # 1.234
List Autoboxing
All the functions from List::Util and select ones from List::MoreUtils
are all available as methods on unblessed arrays and array refs.
first, max, maxstr, min, minstr, minmax, shuffle, reduce, sum, any, all,
none, true, false, uniq and mesh.
The have all been altered to return array refs where applicable in order
to allow chaining.
@array->grep(sub{ $_->is_number })->sum->say;
@array->foreach( func($item) { ... } );
Works like the built in `foreach', calls the code block for each element
of @array passing it into the block.
@array->foreach( func($item) { say $item } ); # print each item
It will pass in as many elements as the code block accepts. This allows
you to iterate through an array 2 at a time, or 3 or 4 or whatever.
my @names = ("Joe", "Smith", "Jim", "Dandy", "Jane", "Lane");
@names->foreach( func($fname, $lname) {
say "Person: $fname $lname";
A normal subroutine with no signature will get one at a time.
If @array is not a multiple of the iteration (for example, @array has 5
elements and you ask 2 at a time) the behavior is currently undefined.
Calculate the difference between two (or more) arrays:
my @a = ( 1, 2, 3 );
my @b = ( 3, 4, 5 );
my @diff_a = @a->diff(\@b) # [ 1, 2 ]
my @diff_b = @b->diff(\@a) # [ 4, 5 ]
Diff returns all elements in array `@a' that are not present in array
`@b'. Item order is not considered: two identical elements in both
arrays will be recognized as such disregarding their index.
[ qw( foo bar ) ]->diff( [ qw( bar foo ) ] ) # empty, they are equal
For comparing more than two arrays:
@a->diff(\@b, \@c, ... )
All comparisons are against the base array (`@a' in this example). The
result will be composed of all those elements that were present in `@a'
and in none other.
It also works with nested data structures; it will traverse them
depth-first to assess whether they are identical or not. For instance:
[ [ 'foo ' ], { bar => 1 } ]->diff([ 'foo' ]) # [ { bar => 1 } ]
In the case of overloaded objects (i.e., DateTime, URI, Path::Class,
etc.), it tries its best to treat them as strings or numbers.
my $uri = URI->new("");
my $uri2 = URI->new("");
[ $uri ]->diff( [ "" ] ); # empty, they are equal
[ $uri ]->diff( [ $uri2 ] ); # empty, they are equal
my @a = (1 .. 10);
my @b = (5 .. 15);
my @intersection = @a->intersect(\@b) # [ 5 .. 10 ];
Performs intersection between arrays, returning those elements that are
present in all of the argument arrays simultaneously.
As with `diff()', it works with any number of arrays, nested data
structures of arbitrary depth, and handles overloaded objects
my @trimmed = @list->trim;
my @trimmed = @list->trim($character_set);
Trim whitespace from each element of an array. Each works just like
their scalar counterpart.
my @trimmed = [ ' foo', 'bar ' ]->ltrim; # [ 'foo', 'bar ' ]
my @trimmed = [ ' foo', 'bar ' ]->rtrim; # [ ' foo', 'bar' ]
my @trimmed = [ ' foo', 'bar ' ]->trim; # [ 'foo', 'bar' ]
As with the scalar trim() methods, they all take an optional
$character_set which will determine what characters should be trimmed.
my @trimmed = ['-> foo <-', '-> bar <-']->trim('-><'); # [' foo ', ' bar ']
Hash Autoboxing
Exchanges values for keys in a hash.
my %things = ( foo => 1, bar => 2, baz => 5 );
my %flipped = %things->flip; # { 1 => foo, 2 => bar, 5 => baz }
If there is more than one occurence of a certain value, any one of the
keys may end up as the value. This is because of the random ordering of
hash keys.
# Could be { 1 => foo }, { 1 => bar }, or { 1 => baz }
{ foo => 1, bar => 1, baz => 1 }->flip;
Because hash references cannot usefully be keys, it will not work on
nested hashes.
{ foo => [ 'bar', 'baz' ] }->flip; # dies
Recursively merge two or more hashes together using Hash::Merge::Simple.
my $a = { a => 1 };
my $b = { b => 2, c => 3 };
$a->merge($b); # { a => 1, b => 2, c => 3 }
For conflicting keys, rightmost precedence is used:
my $a = { a => 1 };
my $b = { a => 100, b => 2};
$a->merge($b); # { a => 100, b => 2 }
$b->merge($a); # { a => 1, b => 2 }
It also works with nested hashes, although it won't attempt to merge
array references or objects. For more information, look at the
Hash::Merge::Simple docs.
my %staff = ( bob => 42, martha => 35, timmy => 23 );
my %promoted = ( timmy => 23 );
%staff->diff(\%promoted); # { bob => 42, martha => 35 }
Returns the key/value pairs present in the first hash that are not
present in the subsequent hash arguments. Otherwise works as
%staff->intersect(\%promoted); # { timmy => 23 }
Returns the key/value pairs that are present simultaneously in all the
hash arguments. Otherwise works as `@array->intersect'.
Code autoboxing
my $sig = $code->signature;
You can query the signature of any code reference defined with `func' or
`method'. See Signature Introspection for details.
If `$code' has a signature, returns an object representing `$code''s
signature. See perl5i::Signature for details. Otherwise it returns
Perl6::Caller causes `caller' to return an object in scalar context.
`die' now always returns an exit code of 255 instead of trying to use
`$!' or `$?' which makes the exit code unpredictable. If you want to
exit with a message and a special exit code, use `warn' then `exit'.
utf8 lets you put UTF8 encoded strings into your source code. This means
UTF8 variable and method names, strings and regexes.
It means strings will be treated as a set of characters rather than a
set of bytes. For example, `length' will return the number of
characters, not the number of bytes.
length("perl5i is MËTÁŁ"); # 15, not 18
`@ARGV' will be read as UTF8.
STDOUT, STDIN, STDERR and all newly opened filehandles will have UTF8
encoding turned on. Consequently, if you want to output raw bytes to a
file, such as outputting an image, you must set `binmode $fh'.
`croak' and `carp' from Carp are always available.
Child provides the `child' function which is a better way to do forking.
`child' creates and starts a child process, and returns an
Child::Link::Proc object which is a better interface for managing the
child process. The only required argument is a codeblock, which is
called in the new process. exit() is automatically called for you after
the codeblock returns.
my $proc = child {
my $parent = shift;
You can also request a pipe for IPC:
my $proc = child {
my $parent = shift;
my $reply = $parent->read();
} pipe => 1;
my $message = $proc->read();
API Overview: (See Child for more information)
$proc_or_parent->write( @MESSAGES )
$proc_or_parent->say( @MESSAGES )
English gives English names to the punctuation variables; for instance,
`<$@'> is also `<$EVAL_ERROR'>. See perlvar for details.
It does not load the regex variables which affect performance.
`$PREMATCH', `$MATCH', and `$POSTMATCH' will not exist. See the `p'
modifier in perlre for a better alternative.
Modern::Perl turns on strict and warnings, enables all the 5.10 features
like `given/when', `say' and `state', and enables C3 method resolution
Provides `CLASS' and `$CLASS' alternatives to `__PACKAGE__'.
File::chdir gives you `$CWD' representing the current working directory
and it's assignable to `chdir'. You can also localize it to safely chdir
inside a scope.
File::stat causes `stat' to return an object in scalar context.
`time', `localtime', and `gmtime' are replaced with DateTime objects.
They will all act like the core functions.
# Sat Jan 10 13:37:04 2004
say scalar gmtime(2**30);
# 2004
say gmtime(2**30)->year;
# 2009 (when this was written)
say time->year;
`gmtime()' and `localtime()' will now safely work with dates beyond the
year 2038 and before 1901. The exact range is not defined, but we
guarantee at least up to 2**47 and back to year 1.
Turns filehandles into objects so you can call methods on them. The
biggest one is `autoflush' rather than mucking around with `$|' and
autodie causes system and file calls which can fail (`open', `system',
and `chdir', for example) to die when they fail. This means you don't
have to put `or die' at the end of every system call, but you do have to
wrap it in an `eval' block if you want to trap the failure.
autodie's default error messages are pretty smart.
All of autodie will be turned on.
autovivification fixes the bug/feature where this:
$hash = {};
Results in `$hash->{key1}' coming into existence. That will no longer
No indirect object syntax
perl5i turns indirect object syntax, ie. `new $obj', into a compile time
error. Indirect object syntax is largely unnecessary and removing it
avoids a number of ambiguous cases where Perl will mistakenly try to
turn a function call into an indirect method call.
See indirect for details.
`want()' generalizes the mechanism of the wantarray function, allowing a
function to determine the context it's being called in. Want
distinguishes not just scalar v. array context, but void, lvalue,
rvalue, boolean, reference context, and more. See perldoc Want for full
Try::Tiny gives support for try/catch blocks as an alternative to `eval
BLOCK'. This allows correct error handling with proper localization of
$@ and a nice syntax layer:
# handle errors with a catch handler
try {
die "foo";
} catch {
warn "caught error: $_";
# just silence errors
try {
die "foo";
See perldoc Try::Tiny for details.
Better load errors
Most of us have learned the meaning of the dreaded "Can't locate
in @INC". Admittedly though, it's not the most helpful of the error
messages. In perl5i we provide a much friendlier error message.
Can't locate My/ in your Perl library. You may need to install it
from CPAN or another repository. Your library paths are:
Indented list of paths, 1 per line...
Command line program
There is a perl5i command line program installed with perl5i (Windows
users get perl5i.bat). This is handy for writing one liners.
perl5i -e 'gmtime->year->say'
And you can use it on the `#!' line.
Some parts are not lexical. Some parts are package scoped.
If you're going to use two versions of perl5i together, we do not
currently recommend having them in the same package.
See for a complete
Please report bugs at or email
perl5i follows the Semantic Versioning policy, In
Versions will be of the form X.Y.Z.
0.Y.Z may change anything at any time.
Incrementing X (ie. 1.2.3 -> 2.0.0) indicates a backwards incompatible
Incrementing Y (ie. 1.2.3 -> 1.3.0) indicates a new feature.
Incrementing Z (ie. 1.2.3 -> 1.2.4) indicates a bug fix or other
internal change.
Inspired by chromatic's Modern::Perl and in particular
I totally didn't come up with the "Perl 5 + i" joke. I think it was
Damian Conway.
Thanks to our contributors: Chas Owens, Darian Patrick, rjbs, chromatic,
Ben Hengst, Bruno Vecchi and anyone else I've forgotten.
Thanks to Flavian and Matt Trout for their signature and Devel::Declare
Thanks to all the CPAN authors upon whom this builds.
Copyright 2009-2010, Michael G Schwern <>
This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.
Repository: Issues/Bugs: IRC: on the #perl5i
channel Mailing List:
Frequently Asked Questions about perl5i: perl5ifaq
Some modules with similar purposes include: Modern::Perl, Common::Sense
For a complete object declaration system, see Moose and MooseX::Declare.