Paul Hampson wrote: > > I think it'd be better to have a separate patch to fix > > sql_finish_query() usage everywhere and not only in > > rlm_sql_postauth() > > SOunds good. Are you going to make it call it always? I expect > such a function would be safe to call at any time... From memory, > the mySQL provides such a function, but its not implemented in > rlm_sql_mysql at the moment. I'll also have to look into that > sometime.
mySQL provides a mysql_free_result() that you're supposed to call after a mysql_store_result(). So I'm thinking that: * sql_select_query() calls mysql_store_result() and therefore sql_finish_select_query() calls mysql_free_result() * sql_query() don't allocate supplementary resources so sql_finish_query() does nothing and it's just fine. Moreover (now that I read the entire source code) if you really want to call sql_finish_query() after a failed sql_query() I think the best place to do it is the function rlm_sql_query() in sql.c and not in a lot of different places in rlm_sql.c Since this issue has nothing to do with rlm_sql_postauth() I think you could safely commit the patch now. -- Nicolas Baradakis - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
