bug#43682: 28.0.50; Clean up nnimap server buffers?

[Robert Pluim]

>>>>> On Mon, 28 Sep 2020 16:37:15 -0700, Eric Abrahamsen 
>>>>> <eric@ericabrahamsen.net> said:

    Eric> Someone noted on gnus.general that their imap connections are 
frequently
    Eric> broken, and they end up with a lot of dead process buffers.

    Eric> I'm talking to them about maybe making the keepalive timeout
    Eric> configurable, but wouldn't also be tidy to clean up dead process
    Eric> buffers? How does the attached patch look?

    Eric> Eric

    Eric> diff --git a/lisp/gnus/nnimap.el b/lisp/gnus/nnimap.el
    Eric> index d797e893f5..7f2ebe279e 100644
    Eric> --- a/lisp/gnus/nnimap.el
    Eric> +++ b/lisp/gnus/nnimap.el
    Eric> @@ -407,8 +407,15 @@ nnimap-keepalive
    Eric>                     (time-subtract
    Eric>                      now
    Eric>                      (nnimap-last-command-time nnimap-object))))
    Eric> -            (ignore-errors              ;E.g. "buffer foo has no 
process".
    Eric> -              (nnimap-send-command "NOOP"))))))))
    Eric> +         (condition-case err
    Eric> +             (process-send-string "NOOP")
    Eric> +           (error
    Eric> +            (if (string-search "has no process" (cdr err))
    Eric> +                (let ((buf (current-buffer)))
    Eric> +                  (setq nnimap-process-buffers
    Eric> +                        (delq buf nnimap-process-buffers))
    Eric> +                  (kill-buffer buf))
    Eric> +              (signal (car err) (cdr err)))))))))))

Thatʼs not how you call process-send-string, and the
nnimap-send-command is there for a reason: it deals with imap sequence
numbers.

Also I donʼt think 'has no process' is the only error
'process-send-string' can signal. I see at least 'not running' and
'Output file descriptor.*is closed'

Robert
--