[time-nuts] Metastability?

[Peter Vince]

Bruce, et al,

     "Metastability" was mentioned again recently - I think I read some
messages earlier this year, but can't remember if they were current, or
in the archive, and can't now quickly find them.  I think it has to do
with latches getting into an undeterminable state when asynchronous
inputs drift through, so not giving a suitable setup time?  I think you
advocate using multi-stage shift-registers to minimise this problem?  I
wonder, if you have a moment, if you could walk me through the problem,
and especially the cure?

          Thank you,

               Peter