On Oct 5, 2007, at 11:35 AM, firstname.lastname@example.org wrote:
> "Erik Huelsmann" <email@example.com> wrote on 10/05/2007 10:07:44 AM:
> > On 10/5/07, Justin Erenkrantz <firstname.lastname@example.org> wrote:
> > > On Oct 5, 2007 12:45 AM, Erik Huelsmann <email@example.com> wrote:
> > > > It's possible to do this, by overloading 'svn cleanup --
> force' to do
> > > > the additional work of the upgrade... I guess the question is
> now how
> > > > the others feel about this. Should the error happen when
> someone wants
> > > > to store a depth or when someone executes a write-operation
> on the
> > > > working copy? (Or even when accessing a working copy for read-
> > > > operations?)
> > >
> > > I don't like confusing the user - any upgrades should be
> > > as they have been in the past. -- justin
> > Ah. But the fact is that transparent updates are actually more
> > confusing, from what I hear from admins... What's confusing about
> > explicit upgrades (more so that transparent ones)?
> How about a config option to enable automatic WC upgrades?
> I would prefer it to be off by default, since we need to mix old
> and new clients (for various reasons) and I don't think most of
> our users will be initially needing the new format.
Not that I have a vote but I would give a +1 to this suggestion. We
got accidentally bitten by an auto-upgrade with 1.4 when we weren't
intending to upgrade. If auto-upgrades were *not* the default it
would be extremely helpful in a multi-project, multi-machine
environment. Even when you try to lock down the version of tools
that people use, occasionally you'll have users which "experiment"
with a newer version. There's no going back once somebody has
accidentally converted their working copy. Doing a fresh checkout
with the old client can be a major pain in certain situations.
> If automatic upgrades are disabled, and the user tries to do an
> operation that needs the new format, it would ask them to upgrade
> the WC.
> Kevin Radke
I would prefer that it just print an error and exit under normal
circumstances. Upgrade of a working copy format really ought to be
an explicit operation. I don't like making it a compile-time-only
option. At a minimum, it should require '--force' to be specified.
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Fri Oct 5 19:28:15 2007