bbackde at googlemail.com wrote: > I know that Berkeley is'nt relational. But according to some > informations I read in the berkeley FAQ you could implement the same > behavior by implementing some of the scan stuff by yourself... Before > I decided to use McKoi I also tried hsqldb (used by thaw) but it was > much slower than McKoi, especially with COUNT queries and some other > queries often used by Frost. According to the hsqldb forums they know > about this issue, but there is still no solution.
I would give Apache Derby a go. Although, I should mention, it has been even slower than HSQLDB for me in some simple tests. The problem with HSQLDB is that its lacking a few key things, such as using indexes for sorts (tablescan is bad, it also makes bigger-than-RAM tables impossible to use in certain ways)! I now see HSQLDB more as a structured in-RAM hashtable than as a real DB. --- John B?ckstrand
