The final status update: PyPy with utf-8 unicodes translates and, with the exception of one last test that is still failing, works correctly (or at least appears to.) Fixing translation issues was easier than I thought it would be, so I most spent the last two weeks fixing bugs that resulted from differences between the behavior of the translated and untranslated code. For example, although RPython supports using __len__ with the built-in len() function, it doesn't use __len__ when evaluating a object as a boolean. As a result, there were a number of locations where the compiled interpreter would seg-fault or hang even though the untranslated code worked correctly.
There's still a few of things left to do in the last week; the aforementioned test being one of them. Unfortunately, I doubt that there will time for my branch to be reviewed and merged before the summer is over, but I think I'll definitely finish the work I have left to do.
No comments:
Post a Comment