Re: timeout due to debug output

[Oswald B. <osw...@gm...>]

On Fri, Jan 27, 2023 at 10:36:19AM +0100, Uwe Kleine-König wrote:
> mbsync -V -DDMns
>
(note that -D implies -V.)

>My interpretation is that some reply is expected from the far side 
>(line 24) so a timeout (default value 20s) is setup. Then mbsync 
>proceeds to load the local box. With the given debugging and mailbox 
>size this takes longer than 20s and then the timeout timer triggers 
>before the connection is checked for input the next time.
>
>Does this sound about right?
>
seems so.

though it's remarkable that printing 75k lines takes almost two minutes.

>I'll increase the timeout (maybe to 0) locally, but a proper fix would 
>be very welcome.
>
i could address the problem somewhat easily by prioritizing socket 
notifiers over (their) timers.
or freezing the timers during local activity. (how ugly.)
however, all these approaches still leave the problem that the remote 
side would ultimately timeout by itself.
fixing this properly would require some form of multi-threading in the 
maildir and proxy drivers, and this would be a major effort.