> We have already discussed this (in fact, you said that you liked the > idea). It would, for example, allow the creation of an email client > which could just ask a server to send a message, rather than having to > wait minutes or even hours to insert the email into the appropriate slot > in Freenet. This should also be usable for "think cash"-style > applications. Basically it transfers the burden of time-consuming > operations from the client to the server.
Yes, but I already have such an e-mail client, which is why I'm asking how this is useful. What I said I liked before was a client API for accessing in-freenet indices. That way my current e-mail daemon could be written in Python instead of Java (it currently uses direct calls to IndexClient). But the index API doesn't necessarily need to be non-blocking as the Python e-mail daemon could be run in the background and connected to by a short-lived client. I think perhaps the answer is that it would allow you to take the part of the e-mail daemon that needs to run in the background and put it in the node and then write the rest of the e-mail code into the short-lived client, eliminating the need for an e-mail daemon as such.
