> I consider my proposal to work better for Nim's stdlib and Nim's compiler
@Araq, I can see how your proposal means minimal changes for stdlib and the compiler, but doesn't it then have some fairly severe limitations? How does your proposal cover the other cases such as shared memory, and doesn't it still have these special cases that won't work and perhaps others that we haven't encountered yet? And isn't it possible that my proposal eliminates many of the limitations while very easily and concisely supporting multi-threading, which the Bacon/Dingle method can't do very well if at all?
