Re: svn commit: r27741 - in trunk/subversion: include libsvn_fs libsvn_fs_base libsvn_fs_fs tests/libsvn_fs

On Fri, 09 Nov 2007, cmpilato@tigris.org wrote:
...
> +static svn_error_t *
> +base_node_origin_rev(svn_revnum_t *revision,
> + svn_fs_root_t *root,
> + const char *path,
> + apr_pool_t *pool)
> +{
> + svn_fs_t *fs = svn_fs_root_fs(root);
> + svn_fs_root_t *curroot = root;
> + apr_pool_t *subpool = svn_pool_create(pool);
> + svn_stringbuf_t *lastpath =
> + svn_stringbuf_create(svn_fs__canonicalize_abspath(path, pool), pool);
> + svn_revnum_t lastrev = SVN_INVALID_REVNUM;
> + const svn_fs_id_t *pred_id;
> + struct id_created_rev_args icr_args;
> + svn_boolean_t found_copies = FALSE;
> +
> + /* Walk the closest-copy chain back to the first copy in our history.
> +
> + NOTE: We merely *assume* that this is faster than walking the
> + predecessor chain, because we *assume* that copies of parent
> + directories happen less often than modifications to a given item. */
> + while (1)
> + {
> + svn_revnum_t currev;
> + const char *curpath = lastpath->data;
> +
> + /* Find the previous location of this object using the
> + closest-copy shortcut. */
> + SVN_ERR(prev_location(&curpath, &currev, fs, curroot, curpath, subpool));
> + if (! curpath)
> + break;
> +
> + /* Update our LASTPATH and LASTREV variables (which survive SUBPOOL). */
> + found_copies = TRUE;
> + svn_stringbuf_set(lastpath, curpath);
> + lastrev = currev;
> +
> + /* Update our CURROOT from our calculated LASTREV. */
> + svn_pool_clear(subpool);

Any reason not to clear the subpool at the beginning of the loop (as per
conventions)?

> + SVN_ERR(svn_fs_revision_root(&curroot, fs, lastrev, subpool));
> + }
...