Re: svn commit: r889320 - in /subversion/trunk/subversion/libsvn_fs_base: obliterate.c obliterate.h revs-txns.c

Julian Foad <julianfoad_at_btopenworld.com> writes:

>> > + 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?
>
> Ah, yes, looks like it should. Thanks. Will check and fix.

Looks like there is a similar problem further on in the same function:

            /* Make a deep copy of the child node-rev. */
            SVN_ERR(svn_fs_base__node_rev_dup(&new_child_id, child_entry->id,
                                              new_txn_id, old_txn_id, trail,
                                              scratch_pool, scratch_pool));

Should that be pasing result_pool as the second pool?

            /* Make the (new) parent node's rep refer to this new child. */
            SVN_ERR(svn_fs_base__dag_set_entry(parent_dag_node, child_name,
                                               new_child_id, new_txn_id,
                                               trail, scratch_pool));

Should that pass result_pool rather than scratch pool?

-- 
Philip