Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 27 Nov 2001 17:01:32 -0500 (EST)
From:      <doug@safeport.com>
To:        Giorgos Keramidas <charon@labs.gr>
Cc:        freebsd-questions@FreeBSD.ORG
Subject:   Re: cvsup question
Message-ID:  <Pine.BSF.4.21.0111271635420.85635-100000@pemaquid.safeport.com>
In-Reply-To: <20011127162503.GA8434@hades.hell.gr>

next in thread | previous in thread | raw e-mail | index | archive | help
I am probably missing something but as I read your example I do not
think that was my point. There should be no timing issue. In my example
all time are Zulu.

Using the cvsup date tag format: at 2001.11.25.15 (or later) I ran a
cvsup with no date tag. I observed that the latest mod I picked up was
2001.11.24.19+. So to get something that (I thought) would be
reproducible I ran another cvsup with a tag of 2001.11.25.00.00.00.
Much earlier than the time of the 2nd cvsup.

I did not get tags anywhere near the boundary times rather I got:

  Add delta 1.291.2.22 2001.11.02.16.45.05 
  Add delta 1.74.2.6 2001.11.23.13.13.19 
  Add delta 1.5 99.08.28.01.01.30
  Add delta 1.5.2.2 2001.07.22.12.40.55
  Add delta 1.4 99.08.28.01.16.04

I do not know how long it takes the mirrors to sync but I suspect it is
less than 5 months ;) and I would hope for better than 2 days. This
analysis lead me to conclude the date tag tests the CVS line in the
file, e.g.: the "* $FreeBSD: <name> <version> <timestamp>" line. Not
when the file actually arrives.

The timestamps of the source files themselves suggest they were added
after my date tag limit, but with CVS timestamps that made them valid to
include. If that is correct then no date tag guarantees a consistent
result. 


On Tue, 27 Nov 2001, Giorgos Keramidas wrote:

> On 2001-11-25 23:06:10, doug wrote:
> > All of the modules listed below were commited in the time between my cvsup
> > runs. The date tag apparently works on what is put in the CVS version line
> > not when the commit is really made. 
> > 
> > So using date=<whatever> does not really mean no new updates will be
> > picked up.
> 
> Yes, this can happen.  If you are user ``X'' sitting at a FreeBSD
> terminal somewhere in the world, and ``S'' is your favorite CVSup
> server, with ``C'' being the cvsup-master which grabs the changes from
> the repository, the following timelines show how this things can
> happen.
> 
> The small letters `a', `b' and `c' indicate deltas that are committed
> and grabbed by ``C'' the cvsup master.  The lines marked with two
> stars at the beginning are points in time that ``X'' fires up CVSup.
[cut]
> As you can see from this simple timeline, the user at ``X'' has
> started his first CVSup session *after* the second delta reached the
> CVSup master server, and asked for all the changes that were older
> than the current date.  Since this particular delta had not reached
> his favorite CVSup mirror, he didn't get to grab it until the second
> time he fired up CVSup though :)
> 
> -giorgos
> 
> 

_____
Douglas Denault
doug@safeport.com
Voice: 301-469-8766
  Fax: 301-469-0601


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0111271635420.85635-100000>