Re: access time changes vs. kevent observer?
Re: access time changes vs. kevent observer?
- Subject: Re: access time changes vs. kevent observer?
- From: Dave Zarzycki <email@hidden>
- Date: Wed, 14 May 2008 12:27:16 -0700
On May 14, 2008, at 9:48 AM, Jim Correia wrote:
On May 14, 2008, at 12:33 PM, Dave Zarzycki wrote:
I filed a bug a few months ago in this regard:
5435816 Add a NOTE_ATIME option to EVFILT_VNODE events
Is your intention to separate out atime from the rest of the
attributes? (That would be perfect.)
Jim,
Yes, although I have no idea when my peers might find the time to work
on the above enhancement request.
For now, you'll just need to ignore some spurious wakeup events.
The current behavior seems wrong at least from the point of view
that it is inconsistent. I'm getting NOTE_ATTRIB for some atime
changes, but not all. What is the expected behavior here today?
(Should I separately file a bug about this, or does your bug above
cover this case?)
I suspect that the NOTE_ATTRIB is accurate. What is probably happening
is that an attribute of the file is changing, but you haven't
discovered which one yet. I don't know why you are observing what you
are, nor do I have the time to look into it. Sorry…
Do you have any advice for how I can detect a spurious atime wakeup
from other meaningful attribute changes? (As time marches on, and
more metadata is added to Mac OS X, trying to cache the last set of
attributes and compare to the current set seems like a bad solution
to this problem.)
Remember the previous values that you're interested in and compare
current values the next time you wake up. If they are all the same,
then you were spuriously woken up.
Good luck,
davez _______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden