On Thu, Nov 12, 2009 at 16:31, Branko Čibej <brane_at_xbc.nu> wrote:
> Greg Stein wrote:
>...
>> I think relative references would not include ".." such that it
>> reaches *out* of the root of the line of development. If you want to
>> do that, then I'd suggest a rooted path like ^/branches/upstream
>> instead.
>
> How do you make a reference to that from trunk, then, unless trunk is
what is "that"?
I'm unclear on why you would ever have a relative reference to
something *outside* of the root of the line-of-development.
> actually branches/trunk? Or if you don't make the references
> branch-relative, how do you move the project around within a larger
> repository, without rewriting all references?
Those absolute ones I referenced... yah. Those are a bugger if you
move the thing around underneath. Unless we had some kind of
"redirect" node.
> The very fact that we're lobbying to move the repository directly to a
> top-level subdir in the ASF repository, instead of going through
> incubator/ first, should raise a red flag -- your rationale for that on
> general_at_incubator implies that Subversion doesn't handle moves all that
> well at all. (Which is nothing new. :) )
No. That is to avoid a copy. Has nothing to do with path handling.
Yes, it *should* be a move, but you can sit around on bust on that
like a grumpy old bastard all you want. It will not change the simple
fact that we do not have moves. Yell and scream "I told you so!" until
you're blue. Or go back in time and propose the solution that you
never did. Any way you want to cut it, we don't have a move. So to
avoid operations that have to cross copy-boundaries (how many?), it is
better for us to simply avoid that copy. And we can.
-g
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2417265
Received on 2009-11-12 22:39:13 CET