The community bonding period is over, that means that the actual Summer of Code begins now. Did I bond enough? Will I get paid? Have I already failed and not noticed? Will the bloody fedex package ever leave Memphis? Only the future can tell, but I'll take the chance to give a sketch of the current project status and what has changed from the original plans.
I've been coding for a few weeks, laying the foundations for the rest of the summer, getting acquainted with the parrot tools and prototyping stuff. I've been taking it easy and moving rather slowly, but the early start it seems to have paid off: Most of the build infrastructure is already in place, the prototyping is basically done, DecNum is starting to take shape, the test harness is in place, and cotto hasn't gotten (visibly) fed up with me bugging him on IRC.
Some design discussions have taken place and prompted some changes but the project schedule is mostly intact and there haven't been any serious setbacks yet. As I mentioned on a previous post, the PMCs are now dynpmcs and not going into the parrot core. After some thought I decided the names would have to be changed as well, to avoid name clashes with the core PMCs we aren't replacing anymore. Instead of being prefixed with "Big" the PMC names will be prefixed "Dec", which ties nicely into an idea I had about putting the emphasis on the 'Decimal' nature of the numbers, rather than on the size (which turned out to be 'not as big as I originally thought' once I started looking at the library internals).
I've also changed my mind on the implementation order, I'll do DecNum and then move on to DecInt, instead of the order I originally suggested on the proposal, I believe things should be both faster and simpler to do this way.
Another goal for the original proposal that I've given some thought to is recycling the current Big* tests, and I don't really think It's worth the effort, It'll probably be better to just convert the decNumber test-suite and use that. Now that compatibility isn't a goal anymore the decNumber tests should be enough to convince us of the correctness of the code.
I think that sums up all the changes to the project so far. It looks to me like we're heading in the right direction, and there might be time for a stretch goal if things go well, but I don't want to get too far ahead of myself. Stay tuned for updates.