This week has seen changes spread across many different pieces of Perl 6. Let’s have a look!
- rurban has gotten the -O2 pbc optimization enabled for parrot and tomorrow’s parrot release will feature that. He measured some 3-5% improvement, but hasn’t benchmarked Rakudo thus far. If nobody beats me to it, I may run some benchmarks this week.
- rurban has also set up a bunch of buildbots, among others for parrot, NQP and Rakudo.
- I’ve missed it last week, but lizmat has been working diligently and effectively on both S11 (modules) and S22 (package format) during these last two weeks!
- At the same time, FROGGS has been moving the integration of PAUSE (the perl programming author’s upload server) with Perl 6 and it sounded like there’ll be a big announcement this or next week regarding that whole complex
- Another thing lizmat did (the week before this one) was improving error messages for when you typo a routine or try to call one that doesn’t exist as well as helpful messages about trying to use .length and .bytes methods on stringy things being banned from Perl 6 (because length and bytes are very ambiguous when applied to a string with regards to Unicode and combining characters and stuff)
- jnthn has advanced NativeCall support on MoarVM to the point where it passes all tests! These changes are already in the master branch of MoarVM, NQP, Rakudo and zavolaj (aka NativeCall)
- jnthn has started the “tristar” branch in the rakudo/star repository to hopefully get a three-backend Rakudo Star distribution this month. Mouq and me helped him a bit. Neither of us three are really good at doing build system stuff, though …
- just as I was writing this, rurban started fixing a nasty regression that made changes to %*ENV not propagate to qx (execute shell commands and return the output as a string) on Rakudo for Parrot.
And here’s a few things people have in the pipe for the near future:
- as mentioned above, FROGGS is still working on ecosystem/cpan/PAUSE/versioning/installation things
- lizmat is likely going to continue work on the S11 and S22 synopses
- I’ve found out that during the CORE.setting compilation, a few strings occur in MoarVM’s generation two more than a few thousand times each. By introducing string interning in the right place, this might bring memory usage down by a significant amount, but my first attempt at string interning just caused mysterious compilation failures.
- there’s quite a bit of momentum towards having tristar finished for this month’s release, so I’m very much looking forward to that
- MoarVM concurrency support is probably going to land after this month’s release, so that it can be tested and debugged sufficiently before it gets unleashed on the world.
- rurban is going to work on a JIT compiler for parrot at some point in the near future
- Mouq has begun a pure-perl GIF decoder as well as a library to read tar files
- hoelzro is likely going to have more time for Perl 6 in the future, so I’m hoping for a complete highlighter for Kate (and thus QtCreator) in a relatively short amount of time
- sadly, there’s a billion incompatible syntax highlighting definition schemes … it would be fantastic if we could generate many highlighters from a common format, but I fear that’ll remain a pipe dream.
Not quite bad progress, considering a big portion of the Perl 6 developers have been slowed down by 2048!
Well, I probably forgot to mention something cool again. I’ll probably put it into next week’s post