On 11/7/05, Robert Schneider <r.schneider@weingartner.com> wrote:
>
> >
> > I am not sure checkout operations are atomic. You do get all files at
> > a particular revision, but the checkout is not all or nothing. If some
> > error happens while checking out, it is aborted and you are left with
> > a partila working copy.
> > And if you are checking out from multiple repositories, it is quite
> > possible that you are not getting the snapshot from all repositories
> > at the very same time.
> >
>
> This is correct. You fetch the source of certain revision. So atomicity
> is not the problem. But would this also mean that if there a more than
> one project in a repository I should use a specific revision when I do
> the check-outs? I must not use HEAD?
>
> I think I didn't understand that right: Why did you mention the possible
> error? Normally I would cancel the build process if something like this
> happens.
>
I was just giving an example when expailing the automicity. The commit
is atomic. That means all or nothing checked in. During the checkin
process if subversion server encounters an error, it will not check in
anything.
Checkout on the other hand will make sure it checkout a particular
revision (possibly HEAD), but it is not automic like commit is atomic.
Say for some arbitrary reason the checkout encounters an error with a
file (locked, or on windows, duplicate file etc..). Then checkout
stops and you now have an incomplete working copy.
If you are checking out all projects using just one command (svn co
<root-url-that-contains-all-projects>) then you can use HEAD, if not
(svn co proj1, svn co proj2 etc) it is better to use a particular
revison number.
--
-Hari Kodungallur
SpikeSource Inc.
http://developer.spikesource.com
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon Nov 7 13:50:24 2005