jkozlowski
commented
9 years ago Jury is out on whether this is actually very useful: on the one hand kdb really is a column database and that is reflected in the IPC format: everything is serialised in column order. That means parsing it out into an object form is always going to involve going through an intermediate representation and the plucking out the values for each row.
That being said, it could potentially be a useful feature for anyone actually using kdb as a dummy object store.
The idea is to have something similar to aeson and associated template haskell deriving machinery.