Re: [Nmh-workers] GCC 8 pre-releases have escaped...

[Ken Hornstein]

>I don't know IMAP, but is there an IMAP server out there that will be
>happy when I overwrite a hunk of ~/mail/inbox/42 in place?  IMAP's a
>textual protocol so we'd be switching one lot of text parsing in scan
>for another.  Would pick(1), when talking to `our' IMAP implementation,
>make use of extensions to the protocol to save it effort, and fallback
>if it's vanilla IMAP?

Good question!

In short:

- Messages in IMAP are immutable, so if you overwrite ~/mail/inbox/42, in
  theory you should get a new UID for that message.  But maybe our "internal"
  IMAP implementation wouldn't care.  I do wonder how UW-IMAP deals with
  that, since it claims to be able to serve up MH folders via IMAP
- The grammar for IMAP is well-defined and pretty compact; I believe you'd
  be able to minimize the client/server I/O.  I don't think a parser would
  be terrible for it.
- AFAICT, pick(1) is implementable completely via the "vanilla" IMAP
  protocol, but if I'm wrong I'd be interested in knowing about it.

--Ken