at the moment, if the TCP session breaks, all outstanding requests will be re-requested. This may not be the wisest thing to do -- instead we could rely on clients (in stub) requesting the same thing another time anyways -- and simplify and make the experience more smooth (remove load on the resolver, and not fire tons of questions when the TCP connection is finally established).
//cc @reynir (and the above was experienced in mirleft IIRC)
at the moment, if the TCP session breaks, all outstanding requests will be re-requested. This may not be the wisest thing to do -- instead we could rely on clients (in stub) requesting the same thing another time anyways -- and simplify and make the experience more smooth (remove load on the resolver, and not fire tons of questions when the TCP connection is finally established).
//cc @reynir (and the above was experienced in mirleft IIRC)