On 12/06/2016 08:41, Arnt Gulbrandsen wrote:
However, the number of aox users is now in two digits and probably
dropping, and they all are capable of adding an extension, so IMO it's
fine either way.
That's a shame.
I wonder if it would be worthwhile considering a 'salvage job' to see
what could be rescued from the codebase?
My understanding is that the message parser goes to great efforts to
handler subtle variations of inputs that are non-standard and that looks
valuable to me if it could be wrapped into a low-dependency library that
does no IO, just passive. (I suspect that actually the test corpus is a
big part of the value, and that's not actually in the source distro is it?)
I'd guess there are similar valuable components in the IMAP state
machine and possibly the local feeds.
I think the database schema is quite valuable too, even though it might
usefully be reviewed for things like the UUID assignment etc and ability
to handle a sharded installation.
I'm wondering whether this sort of thing might enable bits of AOX to
live on, whether supporting a core in Rust or Go or Pony etc, or even
just to rework the application core to C++11 or C++14.
