This is the mail archive of the
kawa@sources.redhat.com
mailing list for the Kawa project.
Re: future model of symbols and environments
Wow - that's a lot to take in. All in all in all, this sounds like a
very nice direction. Two thoughts occur now (probably more later):
- The hot-swapping feature sounds wonderful. So that we can know the
performance cost of using non-static modules we should run some
benchmarks. Does anyone have a good idea for a benchmark?
- The dynamic scoping seems a bit tricky. On the one hand, keeping
the current style and semantics of fluid-let seem (at least to me)
the best fit with a multi-threaded system. It's simple and clear.
On the other hand I have at times wished for a more fine grained
control of how variables are inherited from thread to thread. Like
the difference between InheritableThreadLocal and ThreadLocal. I
have some designs floating around my head, but nothing concrete.
BTW, fluid-let *is* documented.
Regards,
Chris Dean