🚀 go-pugleaf

 > Now here's more info: I just received the I/O error message when my
 > typeahead command had nothing to do with writing to disk -- this time
 > while "S"aving I wanted to see the "p"revious message.  Before this
 > I'd only received the I/O error when I'd typed ahead to do C-x s or
 > something involving another write to disk.

What I _think_ is happening is the typeahead is causing SIGIO to
be sent to Emacs, interrupting the write() system call.  Emacs
expects write() to return exactly the number of bytes that it
told the system to write, and assumes failure if a no all the
bytes are written at once.  What I haven't yet figured out is why
Emacs on a Sun would be using interrupt driven I/O.