Thanks Kai... This is a really weird bug. The old implementation used to build the remapped resultmap in the session scope, but perhaps it got lost somehow and ended up modifying the shared result map. Very strange.
I'll likely end up putting the sync block in there as the solution for now. It will create a bit of a bottleneck, but it's per query, not per row, so it's not too bad. I'll see if there's a way to make it asynchronous and isolated between threads though. Clinton On Sun, Sep 7, 2008 at 3:04 PM, Kai Grabfelder <[EMAIL PROTECTED]> wrote: > Hi there, > > AutoResultMap is not threadsafe if remapResults=true (thus only statements > without an explicit resultMap mapping are affected). > https://issues.apache.org/jira/browse/IBATIS-519 shows the issue. It can > be easily reproduced by using the attached maven project [1] (mvn install > shows the error). I attached a patch to the jira issue. I propose to include > it in the 2.3.4 release. Any objections > > Many thanks > > Regards > > Kai > > [1] > https://issues.apache.org/jira/secure/attachment/12389639/threadtest.zip >
