Let's look at the todo list I published a few days ago:

• Calling UserTalk code from Python: done
• Restore Python "rendering": easier to do now that the previous bit is done, but still not implemented
• Integrating Python exceptions into the Frontier error reporting system: still to be done
• Enabling "KeyboardInterrupt" for runaway Python code: kind of done, but not quite sure how to do this without really disrupting performance
• Convert to using PyGIL* functions for GIL manipulation: I made an attempt, but it didn't really fit the model I'm using right now, and I had to track down a bunch of deadlocks -- since my current method works right now, I'm putting this off until post-1.0
• Flesh out type conversion: This and the error message integration are really what's needing to be finished at this point. They are both not-so-glamorous and yet oh-so-necessary for a polished product. So this is where my focus is right now.

So the 1.0 release is shaping up to be primarily a functionality release, much as it should be. When I start getting real-life complaints about usability, I will have a better idea of where optimization is needed, and will apportion my accordingly. Right now the challenge is to stay focused on shipping, and to get a useful implementation out the door.

I want to emphasize what I mentioned yesterday -- if you really want to play with it, if you want a head start, check out and build the cvs version. I haven't posted any builds yet, but that may change soon.

11:27:02 AM    comments ()  trackback []