Simon Marlow wrote:
> Hmm, good points. The delay before running the signal handler shouldn't be
> too long, unless you have *lots* of running threads.
Actually it is long in the case of the program I'm trying (1 minute and counting).
I think that's because input is done via a _ccall_ (using Readline actually).
Would I be right in thinking that _ccall_'s bring everything else to a stop?
This certainly is a slight disadvantage with the new interrupt mechanism.
But I don't know what's to be done about it, since the idea of running the
interrupt in a separate thread is really a great improvement over the old
mechanism, and I'd hate to see it go.
