Strangely Consistent

Theory, practice, and languages, braided together

Yapsi 2011.02 Released!

It is with extreme...

...hm...

I would like to digress a bit and tell a story. A week ago, I put on my running shoes for the first time in my new hometown. I had already seen out a suitable route on Google Maps, committing the more important street names to memory. It was a cloud-free day, and the sun stood as high in the sky as it ever will in January in Sweden. Out I went.

Before half an hour had passed, I was completely lost. I didn't see any of the streets I had memorized, nor did I run into any of the big roads I knew I would run into if I ran too far.

Gradually, I found myself out on the countryside. That wasn't part of the plan at all. Fields stretched out in all directions. Airplanes criscrossed the sky, their exhaust trails leaving nice patterns behind, reminiscent of some CS books about graph theory.

I started down the country road in the direction back to town, only to have the road slowly curve back in the other direction. It was like one of those text adventure games where you exit one location to the north but end up entering the next location from the northwest! Not conducive to getting somewhere at all.

The the sun went down. At this point, I had been running for over an hour, and was wondering whether I would sleep in a bed that night. I was getting cold and a little bit miserable. The battery of my mp3 player died.

Things got gradually better, though. I found a bigger road, and a sign pointed back to my city, saying it was only five kilometers away. My speed had dropped a bit due to hopelessness, but now it picked up again. I passed a suburb, a mall, a school, a number of unfamiliar blocks, some familiar blocks, and then I was home again. Exhausted. But grateful.

The take-home message is, I hope, crystal clear. A refactor, just like a run, is a process whereby you hope to end up in the same place as you started. Oh, and sunsets can be very pretty.

Anyway.

It is with an exhausted but satisfied feeling that I announce on behalf of the Yapsi development team the February 2011 release of Yapsi, a Perl 6 compiler written in Perl 6.

You can download it here.

Yapsi is implemented in Perl 6. It thus requires a Perl 6 implementation to build and run. This release of Yapsi has been confirmed to work on all releases of Rakudo Star to date. The test files only work flawlessly on Rakudo Star 2011.01, though, due to s/done_testing/done/.

Yapsi is an "official and complete" implementation of Perl 6. This has been confirmed, documented, jokingly referred to, and lamented in a number of places online and offline.

This month's release is a bit late, for which I'm either terribly sorry, or hereby announce that from as of this release, Yapsi will release on the first Saturday of every month. Haven't decided yet.

This month's release could be called a "developer release", but let's not go that far. Suffice it to say that Yapsi behaves the same as last month, but the internals are now much more hackable than last month, so if you've secretly been thinking of becoming a contributor, now's an excellent time to pick some low-hanging fruit. For example, the daughter project 'sigmund', mentioned at the bottom of every Yapsi release announcement, is now feasible; it wasn't really before.

Also, Yapsi has the cutest AST output of all the Perl 6 implementations:

$ bin/yapsi --target=FUTURE -e 'my $a; { $a = 42 }; say $a'
Block -- B0 [$a]
  Var -- $a
  Block -- B2
    Assign
      Var -- $a
      Val -- 42
  Call -- &say
    Var -- $a

It's so cute, it almost looks like Ruby!

Structurally, FUTURE is a re-thinking of PAST. FUTURE is more streamlined than PAST, because it gets rid of the Stmts type. FUTURE is also more versatile than PAST, because it adds 7 new types. In most other respects, FUTURE is like PAST.

For a complete list of changes, see doc/ChangeLog.

Yapsi consists of a compiler and a runtime. The compiler processes a piece of source code, turns it into an annotated tree structure known as FUTURE, and then serializes this tree into a sort of assembler code for a virtual machine. (The virtual machine, being virtual, doesn't really exist. Which, all things considered, is probably a good thing.) The SIC is then... consumed... by the runtime which does its thing and executes it.

With each new release of Yapsi, the old SIC format is thrown out the door, and a new one, sometimes very similar, sometimes identical to the old one, is employed instead. This process is codified for the purpose of keeping people on edge. FUTURE, however, abides by a sofisticated deprecation policy, which in short declares that the format never changes, except in very rare cases when it does.

An overarching goal for making a Perl 6 compiler-and-runtime is to use it as a server for various other projects, which hook in at different steps:

Another overarching goal is to optimize for fun while learning about parsers, compilers, and runtimes.

Have the appropriate amount of fun! \o/