Skip to main content.

Archives

This is the archive for 26 February 2007

Monday, February 26, 2007

I secretly planned to get rid of bison the same way I got rid of flex, one day or another.
However, rethinking about it, Bison seems not to be that bad at all. It's just that I used it a bit forcefully in the early days.

As a result, compilation errors may (supposedly, actually I have no proof of that) cause memory leaks. Some of the "early contextualizations", like i.e. creating a function context after the function declaration and the function name symbol, are a bit clumsy now that I would like to give full error detection support and error context recovery.

But that's not a problem of bison, but a problem of the early falcon compiler code.

I definitely have to fix and reshape function, classes and object declaration.

The rest is not bad, and stands quite gracefully, except, maybe, for dictionary declaration and range vs. array declarations.

But there will be time for that after beta, in the optimization round.