Perl 6 in 2010

2010 has been a busy year for the Perl 6 developers, and came with a noticeable distribution release, many new modules and much fun for the people involved. Here's a short, subjective reflection of this year's Perl 6 events.

Specification

While some specification changes had substantial impact on the compiler writers (and were usually in turn triggered by their worries), the user mostly saw maturing of experimental features, smoothed APIs and some few new features.

Lists, lists, lists

Just as last year, there were a lot of discussions on how lists and related types worked. Much of it was driven by the efforts to implement proper lazy lists in Rakudo.

The result is a much more solid list model, which uses immutable iterators under the hood - a fact that is hidden quite well from the user.

The sequence operator (previously known as "series operator") is a powerful tool for creating lazy lists. It has been extensively refactored to solve problems both with its implementation and usage. It now takes an optional closure left of the ... operator, and a limit that terminates the sequence if it matches true:

# Lazy list of Fibonacci numbers up to (but excluding) 100:
my @fib := 1, 2, *+* ...^ * >= 100;

Date and Time

After several iterations, excessive bikeshedding and serious hacking, Perl 6 now has built-in classes for handling times and dates. They are inspired by the DateTime and Date::Simple Perl 5 modules. The biggest difference is probably that DateTime objects are immutable in Perl 6.

This part of the specification is implemented completely in Rakudo.

Zip meta operator

The Z zip operator can now also act as a meta operator. Thus an easy way to add two lists pairwise and lazily is now

my @sum = @list1 Z+ @list2;

Other changes

• The .pick method, which randomly takes one or more elements from lists and hashes, has been split up into two: @a.pick(3) returns 3 distinct, random items from array @a, while @a.roll(3) does three independent random choices, resulting in possible duplicates in the result list.
• The scoping of lexical multi routines and their protos has been clarified and overhauled (see this discussion of what was wrong previously, and the resolution).
• The numeric roles, buffers and Whatever type have received significant updates

Community

In 2010 we had a remarkable influx of friendly, interested, skilled and enthused newcomers to the Perl 6 community. This is the result of increased marketing outside the Perl community, well publicized releases, great technology and a friendly community.

Community expansion

Two challenges or contests have been announced this year.

Moritz Lenz published a series of "weekly" challenges, guided tasks to implement something that the Perl 6 community needs: A website for the ecosystem, a feature in a compiler and other small things that could be tackled without much prior knowledge.

The overall response was very good, and several people used it as a quick start into the Perl 6 community, and stayed.

Towards the end of the year, Carl Mäsak announced his Perl 6 coding contest. The submitter with the best solutions to five well known programming tasks is to win 100€ worth of books.

Another far-reaching project was the Perl 6 advent calendar for 2010, which attracted more than forty thousand visitors .

In an attempt to make Perl 6 compilers easier available to the masses, John Harrison implemented a web frontend to the Rakudo REPL and made it available at try.rakudo.org.

Conferences

There was a big Perl 6 hackathon (though we had more discussions than hacking) at YAPC::EU 2010 in Pisa. Many Perl 6 contributors, compiler writers and users met and discussed pressing topics in the realms of specification, implementation roadmap, measuring progress and community management. See the meeting notes for details.

Of course there were also some Perl 6 talks at YAPC::EU, many of which seemed well received by the audience.

Perl 6 talks were also held at the Netherlands Perl Workshop, YAPC::Russia, Norwegian Unix User Group and OSDC France, as well as many other conferences which the author forgot :-).

Repository changes

Due to neglected maintenance, the Pugs repository had to be shut down. It has been migrated to git, and split up into several repositories under the perl6 organization on github. Notable parts include:

roast

the Perl 6 test suite

specs

the specification

perl6.org

the main Perl 6 website

modules.perl6.org

the Perl 6 modules website

ecosystem

the module list repository

mu

the remnants of the old pugs repository

While the transition was mostly ad-hoc and not really planned for, most of the resulting confusion could be resolved fairly quickly.

Module ecosystem

While we still lack a proper module distribution system, we now have a website of known Perl 6 modules and a module installer.

But most importantly the number of modules and module authors is steadily increasing (82 known Perl 6 modules at the time of writing, compared to 45 last year). While we still lack the wealth of the Perl 5 ecosystem, there are now database modules, HTTP client and server modules, serialization, file handling tools and so on.

Implementations

Rakudo

Most importantly, this year saw the first Rakudo Star release. Rakudo star is a distribution of the Rakudo compiler, modules and documentation. While it is still a kind of preview release, some few production usages of the Rakudo Perl 6 compiler and distribution have been spotted in the wake of this release.

Also a good part of the Rakudo code based has been replaced during a major refactor, which bases Rakudo on top of a new grammar engine.

Major improvements to the compiler include

• an implementation of lazy lists
• lexical classes and roles
• Perl 6 level stack traces
• much more solid meta object model, which allows the user to create and modify classes programmatically at run time
• implementation of the s/search/replace/ and s[search] = replace() syntactic features, along with several new regex adverbs and variable interpolation into regexes
• improved interpolation in double-quoted strings: array and hash variables now properly interpolate when the expression ends in a bracketing construct
• an improved read-evaluation-print loop, which now remembers variables from previous lines, and also automatically prints the result if no output was produced
• multi level array and hash autovivification
• binding and read-only binding of variables
• a solid implementation of the DateTime and Date classes
• MAIN and USAGE subroutines
• the magic $*ARGFILES file handle and get (comparable to while (<>) { ... } in Perl 5)
• an implementation of basic feed operators

During YAPC::EU the Rakudo contributors decided to target multiple virtual machines: besides the current parrot backend we want to support at least the CLR (.NET).

With this goal in mind, and the need for major performance improvements, Jonathan Worthington prototyped a new, efficient meta object model for parrot in C#, and used that as a base for the new CLR backend. He got help from Matthew Wilson, and Martin Berends started porting the effort to the JVM. Jonathan explained some of his work nicely on the 6guts blog.

In 2011 we will likely see a port of the meta object implementation to parrot, and the beginnings of a Rakudo port to the CLR and JVM.

Niecza

In June, Stefan O'Rear started taking notes on how to compile Perl 6 to the Common Language Runtime (CLR). In November he announced the Niecza Perl 6 compiler, focused on the generation of efficient code.

It already has an impressive list of features, including proper Longest Token Matching, a feature of regexes and grammars that no other Perl 6 compiler has implemented so far.

Summary

2010 was a very rewarding year for the Perl 6 community. With Rakudo there was a compiler available, with which small and medium scale projects can be fun to write. Niecza is quickly catching up.

People experiment with Perl 6, join the community and bring fresh ideas. There is still a long road ahead of us, but the author feels that this road is getting broader and more accessible with each step.