my feelings about torque – an update
my post in october about how great Torque is might be misleading. after finishing raleighlist which uses Torque for persistence i can still say its absolutely pukka but has its problems.
the biggest problem i encountered was a bug in which it fails to retrieve reliable connections from its database pool. it seems like it should be validating the connection but i found it to fail quite often. i just received an email from another user on the turbine mailing list that is experiencing the same issue on tomcat and oracle. my workaround was to hack the torque source code so that it uses resin’s connection pooling. problem solved.
then next problem i encountered was performance. when performing certain selects on tables with a large number of records, the performance degraded quickly. i’m sure there are tweaks i could implement and perhaps i just need to explore proper indexing techniques but instead i just wrote jdbc code in poor performing areas and it now performs extremely well even with 50k records.
