On 10/4/07, Jack Repenning <jrepenning@collab.net> wrote:
> On Oct 4, 2007, at 11:02 AM, Mark Phippard wrote:
>
> >
> > If it comes down to not being able to [heal 1.4 damage on the fly],
> > what do you prefer? An
> > absolute bump or just one when --depth is used?
>
> I would absolutely prefer an absolute bump.
>
> The problem I see with saying "depth is a 1.5 feature and using it
> commits you to 1.5" is that it requires the user to remember every
> command s/he's ever typed, cross-mapped to every working copy s/he
> may have. The problem I see with the situation in general is that
> these corrupted working copies misbehave in ways you might not
> necessarily notice for a while (depending on which subtrees got
> overpopulated), so by the time you notice the misbehavior you may
> long have forgotten that one foolish experiment with "--depth".
I disagree ... let me try to explain.
First, with this plan, there would be no WC corruption. If you used
the --depth feature with a 1.4 WC you would lose the ability to use
1.4 clients on that WC. That might be confusing but it still makes it
no worse than the absolute bump idea and gives users a chance to have
multiple client coexist for a while.
Second, it seems unlikely you are going to accidentally use depth and
then regret it. I do not see why you would need to remember anything.
Worst case scenario is that you try to use a 1.4 client and it tells
you that you need a newer client. I think when you talked about
remembering you were referring to not doing any kind of format bump.
In that scenario, you are right you would have to remember so that you
do not use a 1.4 client and corrupt the WC. That is a different
scenario.
> > I could easily see GUI clients adding a 1.4-compatibility preference
> > that triggered us to just hide the depth option from the dialogs and
> > always use infinite/unknown.
> >
> > I actually do not think it would be that difficult to explain that
> > depth is a 1.5 feature and using it commits you to using 1.5 clients.
> > I hate the idea of an absolute format bump.
>
> Me too. But I hate even more the idea of a lurking, piecemeal, hit
> you when you least expect it, dependent on really hard-to-explain
> abstract features, leaves no evidence, sometimes-the-magic-works
> format bump.
Again, I do not think it would be that confusing. On the other hand,
we know from real experience with 1.3 -> 1.4 that a WC format bump
does cause problems and confusion.
--
Thanks
Mark Phippard
http://markphip.blogspot.com/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Oct 4 20:18:40 2007