julianf...@apache.org writes:
> +svn_error_t *
> +svn_fs_base__node_rev_dup(const svn_fs_id_t **new_id,
> + const svn_fs_id_t *old_id,
> + const char *new_txn_id,
> + const char *old_txn_id,
> + trail_t *trail,
> + apr_pool_t *result_pool,
> + apr_pool_t *scratch_pool)
> +{
> + node_revision_t *noderev;
> +
> + /* We only want to dup a node-rev if it "belongs to" (was created in) the
> + * txn we are replacing. */
> + if (strcmp(svn_fs_base__id_txn_id(old_id), old_txn_id) != 0)
> + {
> + *new_id = old_id;
Should this dup old_id into result pool?
> +/* Create a new representation that is a duplicate of the one keyed by KEY,
> + * but make the duplicate refer to NEW_TXN_ID.
> + * Set *NEW_KEY to the key of the new representation.
> + * Work within TRAIL within FS. */
FS looks like it's a parameter, but it's not.
> +svn_error_t *
> +svn_fs_base__rep_dup(const char **new_key,
> + const char *new_txn_id,
> + const char *key,
> + trail_t *trail,
> + apr_pool_t *pool);
> +
> +/* If the node_rev identified by OLD_ID was not created in transaction
> + * OLD_TXN_ID, then set *NEW_ID to OLD_ID and return. Otherwise:
> + * Make a deep copy of node OLD_ID, with any references to OLD_TXN_ID
> + * replaced by NEW_TXN_ID (### and more differences?) The new node-rev-id is
> + * OLD_ID except with the txn-id field changed to NEW_TXN_ID.
> + * Set *NEW_ID to the new node-rev-id, allocated in RESULT_POOL.
> + * Work within TRAIL within FS.
ditto
> + */
> +svn_error_t *
> +svn_fs_base__node_rev_dup(const svn_fs_id_t **new_id,
> + const svn_fs_id_t *old_id,
> + const char *new_txn_id,
> + const char *old_txn_id,
> + trail_t *trail,
> + apr_pool_t *result_pool,
> + apr_pool_t *scratch_pool);
--
Philip
