On Tue, Oct 16, 2018 at 6:39 PM James Coleman <jtc...@gmail.com> wrote: > Summary: > The new function tuple_data_record() parallels the existing > tuple_data_split() function, but instead of returning a bytea array of raw > attribute heap values, it returns a row type of the relation being examined.
I've been doing something similar with a modified btreefuncs.c/bt_page_items() recently. The approach I've taken is pretty grotty. It would be nice to have a reasonable way of doing this, for both heap tuples and index tuples. (Relatedly, bt_page_items() should probably have a bytea "data" field instead of a text "data" field.) -- Peter Geoghegan
