Re: Reliability of RPC services

[Jonathan S. Shapiro]

On Tue, 2006-04-25 at 12:08 +0200, Marcus Brinkmann wrote:

> So, what Jonathan says is only true at the abstract level for Coyotos:
> The way reply capabilities are implemented using the payload match
> they can be used several times, until the client changes the payload.
> In fact, they can be reused at a later time if the client would change
> back the payload again.

In Coyotos, the difference (in practice) between a send capability and a
receive capability is that the payload in the corresponding FCRB is
automatically incremented as a consequence of invocation.

shap