On Mon, Jul 21, 2008 at 04:51:25PM -0400, Karl Fogel wrote:
> "Rui, Guo" <timmyguo_at_mail.ustc.edu.cn> writes:
> >> What if a child has both entry->deleted==TRUE and
> >> entry->depth==svn_depth_exclude? Or is that supposed to be impossible?
> >
> > We should prevent this from happening. entry->deleted == TRUE just means that
> > the item is gone from the repos. However, the exclude assumes the item exists
> > in the repos.
>
> Just curious: if we have a tree T with excluded item E, what happens if
> we do an update and it turns out E is now removed in the repository? Do
> we locally receive the removal, and therefore remove any record of E in
> the working copy?
Good! I did not consider the situation of receiving the remote removal of an
excluded subtree. I think ideally we should be able to remove the exclude
entry once the server side counterpart is gone. In such a way, we can
confidently raise a conflict when the user is trying to pull in a name that
exists in an exclude entry.
To handle this, we need some cooperations between the server and the client.
The server have to report deletion on excluded path and the update editor
should handle this situation properly. Could you give me a hint on the server
side modification for this purpose? I'll write a test case for it and mark it
as XFail() for now.
Rui
PS: Just a reminder, I have another patch pending (the 'copy' one). I would
like to have it merge back once it's approved.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-07-22 14:04:06 CEST