Tim Rowledge <tim at sumeru.stanford.edu> wrote:
> Ken Causey <ken at ineffable.com> is claimed by the authorities to have written:
> 
> 
> > P.S. I've got to wonder if this couldn't be handled at some relatively
> > low level in a handful of places and remove this bit of policy.  Surely
> > there are a small handful of code pieces that must be passed through
> > before any use of network can be made, such as Socket creation.
> >
> Socket creation probably isn't the place to do it - there must surely be
> situations where you need to create sockets but don't yet want to make
> any attempts at connection.
> 

I dunno, this seems like a marginal usage.  Putting an initializeNetwork
in the two or three #new methods would seem to get the network
initialized in any normal case.  The reasoning: old sockets will be
unusuable across snapshots, and thus any socket usage must begin with
creating a new socket.  The only exception is resolver lookups.


> One possibility is to add an exception signal to the appropriate
> primitive methods and have the default action be to try to init the
> network. Of course, that means one needs to be able to work out the
> reason the prim failed which can be difficult sometimes since we get no
> information except the fact of failure.

Neat idea.  Yet another use for this ability....


-Lex