inotify polling in master branch

Kevin J. McCarthy kevin at 8t8.us
Fri Jun 8 22:50:06 UTC 2018


On Fri, Jun 08, 2018 at 01:51:08PM -0500, Derek Martin wrote:
> On Fri, Jun 08, 2018 at 02:14:46PM +0200, Vincent Lefevre wrote:
> > On 2018-06-05 10:03:55 -0700, Kevin J. McCarthy wrote:
> > > Just a heads up that inotify polling was committed to master earlier
> > > this week.
> > > 
> > > There _are_ still a few quirks with it that I'm looking into.
> > > Specifically weird behavior with Esc-prefixed commands.  I won't have
> > > time to debug that until this weekend though.
> > > 
> > > If you notice other issues, please let me know.
> > 
> > I've just tried it, and there is a major regression: it now fails to
> > detect new mail at all in the current mailbox, even when moving the
> > cursor, viewing messages, modifying the mailbox and synchronizing!
> > 
> > I'm using the Maildir format.

Vincent, so far I can't duplicate this.  I'm a bit surprised, because
the mail detection code itself wasn't changed.  inotify is watching the
maildir and returns a "timeout" event - but Mutt still has uses
$timeout and $mail_check normally.

Can you double check the change in behavior by switching between stable
and master again?

> I haven't looked at the patch, but I have been meaning to point out
> that inotify can miss events, under system load (I think it's mostly
> kernel memory pressure, but exactly how is probably irrelevant).  So
> if Mutt will include such a feature, it would probably be prudent to
> not rely on it solely.

Yes, this is an addition.  In fact, $mail_check is still consulted, and
if you have it set to a large value "mailboxes" will won't be noticed
much faster.

-- 
Kevin J. McCarthy
GPG Fingerprint: 8975 A9B3 3AA3 7910 385C  5308 ADEF 7684 8031 6BDA
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.mutt.org/pipermail/mutt-dev/attachments/20180609/ed76d0d5/attachment.asc>


More information about the Mutt-dev mailing list