Tatamata made great job with zmsql but zmsql can be much more faster at least 2 or 4x.
He copies the results(TjanRecordXYZ) from jansql to tbuffereddataset.
JanSql is very fast, when JanSql can create tbuffeddataset without TjanRecordLists it would be fast.
I am thinking to separate JanSql/zmsql to engine (tdatabase) , ttable, tsql like bde. But I am very busy at moment.
Can someone confirm this ? Might be interesting.