PATCH: Implicit activation

[Havoc Pennington]

Hi,

On Sun, 2004-03-14 at 18:53, Richard Hult wrote:
> I prefer "activate" over "start" to avoid confusion as the term is
> "activate" in the rest of the API. set_auto_activation() sounds good to
> me. I've updated the patch accordingly. 

Here's a question: should auto activation be the default?

I haven't thought about this much.

Regarding the check for autoactivation on the bus driver or NULL, is it
really necessary to check that or could we just ignore it? After all
it's fine to autoactivate a service that's already running.

The patch looks good to me at this point as long as a spec change
documenting the flag and semantics goes in at the same time.
Specifically need to document that the message to the autoactivated
service is blocked until the service activates or fails to activate, and
that in addition to the normal error replies to the blocked message you
can get an activation failure error.

Havoc