currently, if t_relay is called and internal failure happens, for example, tcp connection cannot be established, an internal negative reply is sent and failure_route is not executed. this is bad, because in order to catch these failures, t_relay needs to be called with 0x02 flag, which complicates the script gets by the need to at least partially duplicate failure_route block code in route block.
further complication results from the fact that there is currently no means for module functions to determine if transaction already exists. so if a function is called from route block and it needs to change request-uri, it has not way to figure out if it should rewrite request-uri or append a new branch. so this tracker item may also qualify as a bug report.
Log in to post a comment.