Michael Wood <mwood@its.uct.ac.za> writes:
> I've just found what I was probably thinking of. From
> http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=231982:
> [...]
> + print "+" * 72
> + for timestamp, digest, op, f, r, br, tags, branches in self.deletes:
> + rel_name = relative_name(ctx.cvsroot, f[:-2])
> + repos_path = branch_path(ctx, br) + rel_name
> + print ' deleting %s : %s' % (r, repos_path)
> +
> + if r == "1.1":
> + continue
> + # If the file was initially added on a branch, the first mainline
> + # revision will be marked dead, and thus, attempts to delete it will
> + # fail, since it doesn't really exist.
> + del repo_state[br][rel_name]
> [...]
Yes, that's the fix that's already attached to issue #1302 (waiting to
commit until I have Robert's repro case, see issue for details).
> Then in
> http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgId=215586
> there is this:
> [...]
> @@ -493,7 +524,8 @@
> # If the file was initially added on a branch, the first mainline
> # revision will be marked dead, and thus, attempts to delete it will
> # fail, since it doesn't really exist.
> - if r != '1.1':
> + #if r != '1.1':
> + if fs.check_path(root, repos_path, f_pool) != util.svn_node_none:
> ### need to discriminate between OS paths and FS paths
> fs.delete(root, repos_path, f_pool)
> [...]
That's part of a much larger patch to add a `--incremental' option,
and is a workaround because the original patch causes a problem for
--incremental, not a bugfix per se (in fact, it would unfix David's
bug!)
I think the --incremental support proposed there leaves some edge
cases uncovered; in any case, incrementality is something to add after
we have the basic functionality working and thoroughly tested :-)
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon May 12 17:48:38 2003