From: Jonathan Nieder <jrnieder@gmail.com>
Date: Thu, 04 Nov 2010 05:04:37 GMT
Subject: Re: [PATCH 1/2] Use git_open_noatime when accessing pack data
Message-ID: <20101104050437.GB15018@burratino>
In-Reply-To: <7vvd4ew687.fsf@alter.siamese.dyndns.org>

Junio C Hamano wrote:
> Jonathan Nieder <jrnieder@gmail.com> writes:

>> Judging from these three use cases, readonly open()s to the worktree
>> should indeed use noatime, but open()s of .git/config, say, should
>> not.  Hmm.
>
> Why not, when you are talking about readonly open?

A .git/config frequently accessed by git is being used, so tmpwatch
shouldn't delete it.  (Meanwhile, a worktree frequently accessed by
git is not precious on that account, and letting git touch the atime
there might mask more useful information about when other programs
touched the files.)

But that is all very fuzzy to me.  I tend to mount noatime except on
/usr (for popularity-contest).

I guess I should put it another way.  What if anything does readonly
have to do with O_NOATIME?  Why shouldn't we always use O_NOATIME?
Why should the operating system provide atime at all?