Eclectic Sunday

My wife is away today for a bridal shower, so I've got much of the day to myself for hacking. Here's a quick rundown of some of the things I'm looking to work on today:

1. My mind was brought back to Matrixy this morning because of a post to the mailing list where another developer talked about creating his own Octave clone. I pointed him to Matrixy, and then did a quick update/build myself to verify that what we had there still works. A few changes to Parrot in the past few months broke some tests, so I fixed those up, and then I even managed to close one ticket as well. I'll write up another post this week about the current state of Matrixy, and what the next steps in that project need to be. I don't think I'm going to do much more work on it today.
   
2. GC, again. I put together a proof-of-concept lazy allocator for the new fixed-size allocator that I added last week, and it appears to work alright. Since the fixed-size allocator isn't actually used anywhere, I haven't been able to benchmark anything to say that it actually outperforms the eager allocator, but we can easily switch between the two using a preprocessor macro at compile time, so it's no big deal. I started trying to write up a similar lazy allocator for the primary PObj allocator, but that problem is significantly less trivial and I haven't been able to get it to work yet. I may play with that a little bit more today, and I will definitely post about it sometime this week.
3. The io_cleanups branch is getting old, critically so, and I want to get that branch tied down very very soon. The biggest thing that it adds is proper support for pipes, and those are mostly the brainchild of Infinoid. If I can tie down the last of the test failures today, I can close the branch without waiting for him (either merge into trunk if it's non-destructive, or take a good diff and start a new more up-to-date branch).
4. Tene has been very eager to get working on a select/poll functionality, which is something that I had been planning to implement for AIO but that doesn't strictly need to be done at that time. So, if he's eager and available, I may help him draft up a solution to that problem.

I doubt I'll get all of that done today, but it's nice to have a list of things to work on.