Greg Hudson wrote:
> On Tue, 2005-10-11 at 19:38 +0100, Julian Foad wrote:
>
>>Given that this requires "svn" to gain a concept of "whole working copy", we
>>can make sure that every .svn knows which of its files _and sub-trees_ contains
>>$UpdateRev$ keywords.
>
> You're right, this would have reasonable performance.
>
> It does involve an upwards walk, which is currently unprecedented, and
> it would be somewhat fragile, and would break for elaborate operations
> taking advantage of working copy severability. For example:
>
> I check out the svn tree on my desktop
> I tar up a subtree and move it to my laptop
> I work on my laptop
> I tar up the resulting working copy and move it back to my desktop
> I overwrite my desktop's subtree with the tarfile
Severing a working copy is one thing.
Joining a severed and modified WC back into its parent by cut-and-paste ... is
that something we expect and intend to work in general? I suppose it currently
should work, and is, to some extent, wanted. Oh. If so, then I think I give
up on the possibility of doing the whole-WC thing efficiently, or at least it
would involve changes to the way people are allowed to work with severed WCs.
Fortunately your simpler vision of an "UpdateRev" keyword avoids this, so
hopefully we don't need to pursue the "whole WC" vision further.
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Oct 11 21:59:09 2005