NAME List::Keywords - a selection of list utility keywords SYNOPSIS use List::Keywords 'any'; my @boxes = ...; if( any { $_->size > 100 } @boxes ) { say "There are some large boxes here"; } DESCRIPTION This module provides keywords that behave (almost) identically to familiar functions from List::Util, but implemented as keyword plugins instead of functions. As a result these run more efficiently, especially in small code cases. Blocks vs Anonymous Subs In the description above the word "almost" refers to the fact that as this module provides true keywords, the code blocks to them can be parsed as true blocks rather than anonymous functions. As a result, both caller and return will behave rather differently here. For example, use List::Keywords 'any'; sub func { any { say "My caller is ", caller; return "ret" } 1, 2, 3; say "This is never printed"; } Here, the caller will see func as its caller, and the return statement makes the entire containing function return, so the second line is never printed. The same example written using List::Util will instead print the List::Util::any function as being the caller, before making just that one item return the value, then the message on the second line is printed as normal. In regular operation where the code is just performing some test on each item, and does not make use of caller or return, this should not cause any noticable differences. Lexical Variable Syntax Newly added in version 0.09 many of the functions in this module support a new syntax idea that may be added to Perl core eventually, whereby a lexical variable can be declared before the code block. In that case, this lexical variable takes the place of the global $_ for the purpose of carrying values from the input list. This syntax is currently under discussion for Perl's map and grep blocks, and may be added in a future release of Perl. https://github.com/Perl/RFCs/pull/33 Aliasing and Modification Each time the block code is executed, the global $_ or the lexical variable being used is aliased to an element of the input list (in the same way as it would be for perl's map or foreach loops, for example). If the block attempts to modify the value of this variable, such modifications are visible in the input list. You almost certainly want to avoid doing this. For example: my @numbers = ...; my $x = first my $x { $x++ > 10 } @numbers; This will modify values in the @numbers array, but due to the short-circuit nature of first, will only have modified values up to the selected element by the time it returns. This will likely confuse later uses of the input array. Additionally, the result of first is also aliased to the input list, much as it is for core perl's grep. This may mean that values passed in to other functions have an ability to mutate at a distance. For example: func( first { ... } @numbers ); Here, the invoked func() may be able to modify the @numbers array, for example by modifying its own @_ array. Performance The following example demonstrates a simple case and shows how the performance differs. my @nums = (1 .. 100); my $ret = any { $_ > 50 } @nums; When run for 5 seconds each, the following results were obtained on my machine: List::Util::any 648083/s List::Keyword/any 816135/s The List::Keyword version here ran 26% faster. KEYWORDS first $val = first { CODE } LIST Since verison 0.03. Repeatedly calls the block of code, with $_ locally set to successive values from the given list. Returns the value and stops at the first item to make the block yield a true value. If no such item exists, returns undef. $val = first my $var { CODE } LIST Since version 0.09. Optionally the code block can be prefixed with a lexical variable declaration. In this case, that variable will contain each value from the list, and the global $_ will remain untouched. any $bool = any { CODE } LIST Repeatedly calls the block of code, with $_ locally set to successive values from the given list. Returns true and stops at the first item to make the block yield a true value. If no such item exists, returns false. $val = any my $var { CODE } LIST Since version 0.09. Uses the lexical variable instead of global $_, similar to "first". all $bool = all { CODE } LIST Repeatedly calls the block of code, with $_ locally set to successive values from the given list. Returns false and stops at the first item to make the block yield a false value. If no such item exists, returns true. $val = all my $var { CODE } LIST Since version 0.09. Uses the lexical variable instead of global $_, similar to "first". none notall $bool = none { CODE } LIST $bool = notall { CODE } LISt Since verison 0.03. Same as "any" and "all" but with the return value inverted. $val = none my $var { CODE } LIST $val = notall my $var { CODE } LIST Since version 0.09. Uses the lexical variable instead of global $_, similar to "first". reduce $final = reduce { CODE } INITIAL, LIST Since verison 0.05. Repeatedly calls a block of code, using the $a package lexical as an accumulator and setting $b to each successive value from the list in turn. The first value of the list sets the initial value of the accumulator, and each returned result from the code block gives its new value. The final value of the accumulator is returned. reductions @partials = reductions { CODE } INITIAL, LIST Since version 0.06. Similar to reduce, but returns a full list of all the partial results of every invocation, beginning with the initial value itself and ending with the final result. N-AT-A-TIME FUNCTIONS The following two functions are a further experiment to try out n-at-a-time lexical variable support on the core grep and map operators. They are differently named, because keyword plugins cannot replace existing core keywords, only add new ones. ngrep @values = ngrep my ($var1, $var2, ...) { CODE } LIST $values = ngrep my ($var1, $var2, ...) { CODE } LIST Since version 0.10. A variation on core's grep, which uses lexical variable syntax to request a number of items at once. The input list is broken into bundles sized according to the number of variables declared. The block of code is called in scalar context with the variables set to each corresponding bundle of values, and the bundles for which the block returned true are saved for the resulting list. In scalar context, returns the number of values that would have been present in the resulting list (i.e. this is not the same as the number of times the block returned true). nmap @results = nmap my ($var1, $var2, ...) { CODE } LIST $results = nmap my ($var1, $var2, ...) { CODE } LIST Since version 0.10. A variation on core's map, which uses lexical variable syntax to request a number of items at once. The input list is broken into bundles sized according to the number of variables declared. The block of code is called in list context with the variables set to each corresponding bundle of values, and the results of the block from each bundle are concatenated together to form the result list. In scalar context, returns the number of values that would have been present in the resulting list. TODO More functions from List::Util: pairfirst pairgrep pairmap Maybe also consider some from List::UtilsBy. ACKNOWLEDGEMENTS With thanks to Matthew Horsfall (alh) for much assistance with performance optimizations. AUTHOR Paul Evans