Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 28 Sep 1997 10:54:11 +0930
From:      Greg Lehey <grog@lemis.com>
To:        Nate Williams <nate@mt.sri.com>
Cc:        Chuck Robey <chuckr@glue.umd.edu>, Richard Wackerbarth <rkw@dataplex.net>, "Jordan K. Hubbard" <jkh@time.cdrom.com>, hackers@FreeBSD.ORG
Subject:   Re: How do I check out a snapshot?
Message-ID:  <19970928105411.00690@lemis.com>
In-Reply-To: <199709272247.QAA28003@rocky.mt.sri.com>; from Nate Williams on Sat, Sep 27, 1997 at 04:47:59PM -0600
References:  <199709272233.QAA27878@rocky.mt.sri.com> <Pine.BSF.3.96.970928063923.313J-100000@Journey2.mat.net> <199709272247.QAA28003@rocky.mt.sri.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Sep 27, 1997 at 04:47:59PM -0600, Nate Williams wrote:
> Chuck Robey writes:
>>>> Suppose that the snapshots are always generated by using a CVS checkout
>>>> that never references the head directly. In other words, in a manner
>>>> that others can reproduce at a later time. (For example, a checkout
>>>> by date, using yesterday's date)
>>>
>>> Except that means lots more work for the release engineer.  The RE has
>>> enough problems just making a release, making sure that the code is
>>> 'synchronized' to a certain date would make it that much harder.  I
>>> think it would be 'easier' to simply tag the tree.
>>
>> I think you mistake his intent, Nate.  I think he meant it this way.  The
>> RE updates his tree, writes down the date/time that it's updated to.
>
> But, the tree isn't exactly in sync. with the 'real' CVS tree, so the
> code may not be exactly the same.  With 'real' releases, the actual CVS
> tree is locked down by the RE, so his tree will be the same, since
> no-one is allowed to modify it.  This isn't the case with SNAPS.
>
> So, the time/date he uses in his private tree may have other changes
> that occur when people commit code to the 'real' tree.
>
> The whole FreeBSD process is a multi-threaded/SMP kind of thing, so you
> gotta think about race conditions and all those other OS kinds of
> things. :)

I still don't see the problem.  CVS already has the solution: the RE
selects a certain date for the snap, basically whatever he wants.  He
waits for the time to pass, pulls a copy of the repository (though
this is not really necessary), and then checks out -current as of the
time he decided for the snap.  Anybody else who has a tree at least
that up-to-date can do exactly the same thing.

Greg




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970928105411.00690>