Re: [Issue 533] New - implement reserved checkouts

Forgive my almost certain misunderstanding of this...

On Tue, 2002-08-13 at 01:24, Branko Èibej wrote:
> The thing I still haven't cleared up in my head is this: if the same
> node is visible as A/B and X/Y, and you "svn lock" A/B, does that mean
> X/Y is locked, too? Both models have their pros and cons. Have to think
> a bit deeper about that.

My understanding:

If X/Y is a copy of A/B, and some changes are made to A/B, then the same
changes are not made to X/Y, since the SVN FS doesn't have "linking"
(alternative names of the same node), but only "copying" (generating a
new, independant, node), so:

svn copy A/B X/Y
vi A/B
svn commit
# A/B is now different to X/Y.

My conclusion, based on above:

Therefore a lock on A/B has no effect on X/Y, since they are different
nodes, albeit they share the same history.

If we created a "link" mechanism, such that:

svn link A/B X/Y
# Node "X/Y" points to the same data as "A/B" in all respects.
vi A/B
svn commit
# A/B is *still* the same as X/Y.

Now, a lock on A/B would have to imply a lock on X/Y, since they are, in
effect, the same node.

This is much the same as on UNIX-alikes, where file locking calls lock
the "open file", not the filename.

> A related question is, does locking affect bubble-up directories? In
> other words, does locking the root of the repository prevent any changes
> to the repo, or just modifying entries in the root?

As for whether locks are recursive, does that not depend on whether we
say:

svn lock --recursive .

or:

svn lock .

Or am I being a bit simplistic here?

Dave.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Aug 13 11:37:47 2002