Perl 6 Tidings from April 2009

Specification

Recently I wrote an article about Perl 6, claiming that the core of the language is rather stable these days. Maybe I was wrong...

• Daniel Ruoso continued some work on the Iterator API
• Result objects are now called "abstract objects" and are available via $/.ast. Larry also incorporated an older proposal to make the captures accessible in the order they appeared in the input text, which doesn't necessarily correlate with their order in which they appear in some capture lists/hashes. The method is called $/.caps. There's also $/.chunks which returns not only captures, but also non-captured-but-matched strings interleaved with captures.
• Larry (de?)promoted junctions to be a native type, and also made the eigenstates method (to access the values stored in the junction) private. This started many discussion on IRC and p6l about what are junctions good for, why don't they cook your lunch for free and if they are really that hard to understand (no, they aren't; but they do require a bit of effort).
  Independently, but also in a strive to make junctions more intuitive, block parameters now default to Object, so they are transparent to junctions. That means that for @array -> $x { ... } won't autothread over $x if there are junctions stored in @array.
• Carl Mäsak introduced a nice syntax for doing smartmatching in multi subs/methods - but didn't quite achieve what he wanted. As a consequence Larry tweaked the specs for type narrowness in multi dispatch, and scared the hell out of Jonathan Worthington who had implemented most of the old dispatch semantics in Rakudo already, and saw his design and works (as well as performance) turning into ping logic clouds. So Larry had mercy, and reverted the worst parts (or marked them as post-6.0 at least).
• The want() function seem to be fairly unimplementable, so I proposed to dispose of it.

Implementations

Rakudo

The March release of Rakudo brought many cool new features (like state variables), and lifted the long standing limitation that outer variables couldn't be accessed within eval().

After the release Patrick also added rx/.../ quoting and (limited but substantial) support for Perl 5 regexes.

Vasily "bacek" Chekalkin contributed a socket implementation to Parrot, which inspired some people to write HTTP clients in Rakudo already. YaY.

Not so exciting but very useful for end users is the proper importing and exporting of subs, implemented by Jonathan.

SMOP

The smop developers are currently working on a refactoring branch for smop which includes better modularization and fewer problems for bootstrapping due to reduce interdependencies. (Don't know if I understood that correctly).

Summer of Code

Sorry, can't tell you all that much. Just that The Perl Foundation received (among others) four Perl 6 or parrot related proposals (and the application is still open), all of which look very high quality and promising to me.

In a month we'll all know more about them, and about which will be funded.