Sat Jul 21 10:57:54 EDT 2018

signal/next and memory/nextMemory

For Seq, does it make sense to get rid of signal and next?
No.  Keep it.

So for generalizing closeProcess, it it possible to implement a
signal/next pair for that?

Here's an idea:

  - only add memory/nextMemory to Seq
  - for emu, keep the more general process/nextProcess

In the current implementation, it is not necessary to initialize the
processes.  So let's propagate that.

Ok this is a bit simpler.  Next: what should it look like at the MyHDL
side?  A memory can be modeled as a register with a specific update

I think this would be it:

(m, rd) <- memory
updateMemory m (we, wa, wd, ra)