> -----Original Message-----
> From: Matthew Bentham [mailto:mj...@artvps.com]
> Sent: dinsdag 16 februari 2010 12:03
> To: Philip Martin
> Cc: dev@subversion.apache.org
> Subject: Re: [PATCH] wc-ng: remove a use of svn_wc_entry_t from
> libsvn_client
>
> Philip Martin wrote:
> > lock_tests.py 13 is failing:
>
> Darn, I thought I had fixed that problem :-) But I now realise I was
> confused between "lock_tests.py" and "locks-test.exe" (the latter
> doesn't even test libsvn_client). A complete 'make check' is impossible
> for me on cygwin, it takes about 4hrs to get to 'merge_tests.py
> [54/75]' and that never seems to complete (I left it running over night
> but no luck so far).
>
> > I think the problem is that svn_wc__node_get_url doesn't work for
> > deleted nodes, it certainly has no code to scan the working tree for
> > deletions.
>
> Hmm, I'm not sure, it looks to me like svn_wc__maybe_get_entry should
> have returned NULL for deleted nodes too, because although it will find
> them, the 'show_hidden' parameter is false, and svn_wc__entry_is_hidden
> is true when entry->deleted.
a deleted entry and a node with entry->deleted set are two different things.
A deleted entry has a schedule of 'deleted'.
When entry->deleted is set, the entry exists in the revision of the parent
entry, but not in its own revision. (in WC-NG this maps to BASE_NODE
'not-present')
Bert Huijben
