On Aug 14, 2009, at 12:32 PM, Daniel Näslund wrote:
> make check failed for:
> depth_tests.py 32
> switch_tests.py 18
> switch_tests.py 21
> Since those were failing on trunk, I'd say we're safe.
> I was looking at svn_wc_walk_entries3. It internally uses
> svn_wc_entries_read so I decided to change that first.
> The doc string for svn_wc_entries_read2 talks about parsing the
> file for adm_access. I'm not sure what it should say. Since the
> file is going to be replaced (is that already done?) with a db it
> probably should say something else.
> I was asking gstein about the caching of entries and he said that it
> handled low level and that I didn't have to deal with it.
Haven't looked at the patch, but I've got a couple of meta questions/
> Use svn_wc_context_t in svn_wc_entries_read2.
> * subversion/include/svn_wc.h
> (svn_wc_entries_read2): New.
> (svn_wc_entries_read): Deprecate.
If we do introduce a new (temporary) entries reading API, it should be
private, live in include/private/svn_wc_private.h, and probably be
named svn_wc__entries_read(). Such an API definitely won't see the
light of a public release, so there's no need to put it in the public
API file (from which it will disappear post haste).
> * subversion/libsvn_wc/entries.c
> (svn_wc_entries_read2): New. Use absolute paths. Add svn_wc_context_t
> * subversion/libsvn_wc/deprecated.c
> (prune_deleted): Copy from entries.c.
> (svn_wc_entries_read): Reimplement as a wrapper.
That being said, we've already got svn_wc__node_get_children(), which
can be used to fetch the absolute paths of a node's children. These
absolute paths can then be used to get individual entries (the current
replacement path) or directly fetch desired information (using APIs
which haven't yet been written).
I think that svn_wc__node_get_children() should probably be used in
favor of svn_wc_entries_read(), but I could be wrong. What do you
Received on 2009-08-14 19:53:18 CEST