We could, if we're going to do anything at all in this area. Another possibility would be to provide a second optional parameter for json{b}_strip_nulls. That's probably a better way to go.
Here's a patch that adds that argument (only for jsonb; no json implementation yet)
I think it looks sane. We're not stripping a top level null, which is one thing I looked out for.
I think we need a json implementation as well, though.
Thanks for having a Look, Andrew;
if there aren’t any other objections, I’ll come back with a json implementation too.
Attached is a v2 patch with the missing json implementation.
