perl5i - Fix as much of Perl 5 as possible in one pragma

  use perl5i::2;

  or

  $ perl5i your_script.pl

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 (http://github.com/schwern/perl5i/issues) or make a fork (forking on github is like a branch you control) and implement it yourself.

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 version.

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.

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.

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.

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 conveniences.

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 http://github.com/schwern/perl5i/issues/labels/syntax#issue/19 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 work:

    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.

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;

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 $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
    $module->require->import;

    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/Bar.pm"->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/Bar.pm"

    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 grouping".

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

separator

The character used to separate groups. Defaults to "thousands_sep" in your locale or "," if your locale doesn't specify.

decimal_point

The decimal point character. Defaults to "decimal_point" in your locale or "." if your locale does not specify.

grouping

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 complicated.

currency

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

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("http://www.perl.com");
    my $uri2 = URI->new("http://www.perl.com");

    [ $uri ]->diff( [ "http://www.perl.com" ] ); # 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 graciously.

    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 ']

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 @array->diff.

    %staff->intersect(\%promoted); # { timmy => 23 }

Returns the key/value pairs that are present simultaneously in all the hash arguments. Otherwise works as @array->intersect.

    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 nothing.

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.

    my($stdout, $stderr) = capture { ... } %options;
    my $stdout = capture { ... } %options;

capture() lets you capture all output to STDOUT and STDERR in any block of code.

    # $out = "Hello"
    # $err = "Bye"
    my($out, $err) = capture {
        print "Hello";
        print STDERR "Bye";
    };

If called in scalar context, it will only return STDOUT and silence STDERR.

    # $out = "Hello"
    my $out = capture {
        print "Hello";
        warn "oh god";
    };

capture takes some options.

tee

tee will cause output to be captured yet still printed.

    my $out = capture { print "Hi" } tee => 1;

merge

merge will merge STDOUT and STDERR into one variable.

    # $out = "HiBye"
    my $out = capture {
        print "Hi";
        print STDERR "Bye";
    } merge => 1;

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;

        $parent->say("Message");
        my $reply = $parent->read();

        ...
    } pipe => 1;

    my $message = $proc->read();
    $proc->say("reply");

API Overview: (See Child for more information)

$proc->is_complete()

$proc->wait()

$proc->kill($SIG)

$proc->pid()

$proc->exit_status()

$parent->pid()

$parent->detach()

$proc_or_parent->read()

$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 order.

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 select.

    $fh->autoflush(1);

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 = {};
    $hash->{key1}{key2};

Results in $hash->{key1} coming into existence. That will no longer happen.

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 details.

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.

You no longer have to put a true value at the end of a module which uses perl5i.

Most of us have learned the meaning of the dreaded "Can't locate Foo.pm in @INC". Admittedly though, it's not the most helpful of the error messages. In perl5i we provide a much friendlier error message.

Example:

    Can't locate My/Module.pm 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...

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.

    #!/usr/bin/perl5i

    gmtime->year->say;

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 http://github.com/schwern/perl5i/issues/labels/bug for a complete list.

Please report bugs at http://github.com/schwern/perl5i/issues/ or email mailto:perl5i@googlegroups.com.

perl5i follows the Semantic Versioning policy, http://semver.org. In short...

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 change.

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 http://www.modernperlbooks.com/mt/2009/04/ugly-perl-a-lesson-in-the-importance-of-language-design.html.

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 work.

Thanks to all the CPAN authors upon whom this builds.

Copyright 2009-2010, Michael G Schwern <schwern@pobox.com>

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

See http://dev.perl.org/licenses/artistic.html

Repository: http://github.com/schwern/perl5i/tree/master Issues/Bugs: http://github.com/schwern/perl5i/issues IRC: irc.perl.org on the #perl5i channel Mailing List: http://groups.google.com/group/perl5i/

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.