On Mar 28, 2012, at 01:54 PM, Stephen J. Turnbull wrote:

Indeed. I think Barry misspoke here. But remember, we're barely out of alpha test, and we don't actually have a standard archiver, just a simple handler to support further development.

I just want to be careful about terminology here. "Handlers" specifically are modules that are invoked during pipeline processing. There's a 'to-archive' handler but all that does is put a copy of the message and metadata into the archive queue.

There are also implementations of the IArchiver interface, which is where the actual work of getting the message injected into the archiver happens. Implementations are free to do whatever they want here, and in fact you'll see wide variety in the existing ones (send to an address, call a subproc, drop a file in a maildir).

The archive queue runner is what loops over the available and enabled IArchiver implementations. So it goes something like:

pipeline queue runner --to-archive-> archive queue runner --IArchivers--+ handler ^-------------|

Cheers, -Barry