Blogs

GSOC NCI Updates

The NCI updates using the libffi are coming along. I ran into a bit of an annoyance when I found out that there are places internally where signatures are not specified for some functions, but luckily it was easy to remedy once I had figured out the problem. All of those cases used the same signature ("vJP"), it was only hard to find the right place to add that.

Adding Optimizations to HLL Compilers with PAST::Pattern

Parrot Abstract Syntax Trees(PASTs) are one of the forms code being compiled in a PCT-based compiler takes before being evaluated or compiled to bytecode. With PAST::Pattern, you can find all the sub-trees of a PAST that match a certain pattern and transform them. This can be used to perform optimizations at the PAST level.

It's Finally Time to Write Some Optimizations (Almost)

Wherein I dash your hopes of spending the weekend hacking on making your favorite HLL compiler generate better code only to give you hope at the end that it will not be long at all before you can:

The Last Week and a Half in PAST Optimization

I noticed yesterday that I forgot to post a blog post last week. I'll try to make up for that with double the blog post goodness this week(expect a second post Thursday or Friday).

Threads are continuations

Parrot has Continuations. In fact, Parrot is *based* on Continuations: rather than having a call stack and stack frames, each sub call has a return continuation - a pointer back to the call frame and bytecode position to return to after the sub completes.

As the Scheme and Ruby users keep telling us, continuations are pretty neat stuff. In fact they're so neat that they give us light weight cooperative threads "for free".

NFG is (somewhere in the vincinity of...) here

"N̈" (Or "n̈" if you don't like caps) is a grapheme from several minor extended Latin alphabets. It occurs in the orthographies of the Jacaltec Mayan dialect, Cape Verdean Creole, and in the rockumentary "This is Spın̈al Tap". Today I want to talk about the injustices this symbol has faced in the past and how, starting today, parrot can right them.

PAST::Transformer and the Foundation for PAST Optimization

Coding for GSoC officially began this Monday, but I decided to get a head start and started working on PAST::Walker Thursday. PAST::Walker is the foundation for the rest of what I will be doing regarding PAST Optimization, and the foundation is largely laid. There have been a few complications, and a few decisions remain to be made(and implemented) before I can say that PAST::Walker and PAST::Transformer are finished, but implementing PAST::Walker has gone smoothly and quickly for the most part.

Dynamic code points, the grapheme tables and not getting your services denied.

One of the features of NFG I've mentioned before is that it solves our problems with variable width characters without taking additional storage space over UCS-4, on top of which it's defined. The artifact that allows us to pull off that trick is called the 'grapheme table' and today, I'll try to explain how it works.

How do threads fit into parrot?

(This post is to show my thinking on my GSoC project. Sorry in advance if it's not especially coherent or interesting out of context.)

Over the past week I've been examining the Parrot code to figure out how hybrid threads are going to fit in. First, if I'm going to make Parrot do stuff concurrently, I have to figure out what Parrot is actually doing at all.

GSoC Project: NCI and Stack Frame Improvements

My Google Summer of Code (2010) objectives are to integrate libffi into the NCI framework and to build a new Stack Frame builder that takes advantage of the llvm.

Syndicate content