[BUG] DB closed to early in 6fb5700

[Nicolas Sebrecht]

On Tue, May 10, 2016 at 12:56:10AM +0200, Nicolas Sebrecht wrote:
> On Mon, May 09, 2016 at 07:50:39PM +0200, Łukasz Żarnowiecki wrote:
> 
> > Hello again,
> > 
> > While I was testing my Python3 branch atop next I noticed that the
> > commit 6fb5700 introduces bug when running offlineimap without "-o"
> > switch.  On second loop iteration it complains that the database was
> > closed so no operations could be performed on it.
> > 
> > I did not look into the code yet (lack of time, sigh), but it seems like
> > a database is closed to early.
> 
> This is possible we close the database too early. I have no idea which
> new iteration of a loop would require it, though.

I could reproduce this bug with autorefresh enabled.

-- 
Nicolas Sebrecht