Victor Engmark wrote:
I would distinguish a "patch" from an "extension" by looking at the
functionality and need for it:
- a "patch" does not provide new functionality, but is needed for the
system to be able to fulfill one or more basic requirements;
- an "extension" provides new functionality, but is not needed for the
system to fulfill the basic requirements.
So to permit "extension" without "patches", the basic requirements,
including extensibility, need to be designed-in from the start. It
is desirable that extensions should be backwards-compatible for
interoperability (i.e. if one party in the mail exchange does not
support a particular extension, the exchange should not fail). That
seems to imply negotiation of extensions with fallback. And that
probably precludes batch-oriented (command only, as opposed to
command-response interactivity) transfer.