From: Jason Sewall <jasonsewall@gmail.com>
Date: Tue, 24 Jul 2007 15:31:07 GMT
Subject: Re: [PATCH] Add a 1-second sleep to git-cvsexportcommit test
Message-ID: <31e9dd080707240831u1625ea70m629ea325f2101e8b@mail.gmail.com>
In-Reply-To: <200707241457.53414.robin.rosenberg.lists@dewire.com>

Fedora 7
Kernel 2.6.22 (SMP)
CVS 1.11.22
ext3 (unsure of version or how to discover version)

As one would expect from the above, the 'granularity' test you gave my
reveals 1-second granularity.

On 7/24/07, Robin Rosenberg <robin.rosenberg.lists@dewire.com> wrote:
> tisdag 24 juli 2007 skrev Junio C Hamano:
> > > Here CVS sleeps. The amount varies between invocations since it
> > > only sleeps enough for the seconds to wrap.
> >
> > Makes one wonder what it would do if you are on a filesystem
> > with coarser-than-a-second timestamp resolution.
>
> Like fat, but then the last test fails on FAT, which wasn't the case. Any other reasonable file
> systems that comes to your mind?
>
> Jason, could you provide us with some more information on OS, fs, cvs version etc.
>
> Whether timestamp granularity is larger than a second or not can be checked with this line, I think:
>
>         touch a && ls --full-time a && sleep 1 && touch a && ls --full-time a
>
> Sample output where the timestamps are roughly one second apart.
> -rw-r--r-- 1 me me 0 2007-07-24 14:15:47.330927250 +0200 a
> -rw-r--r-- 1 me me 0 2007-07-24 14:15:48.338990250 +0200 a
>
> -- robin
>