On 2022-Apr-18, Tom Lane wrote:
> I suppose that Peter was trying to remove special cases from the
> outfuncs.c code, but do we want to put this one back? Richard's
> proposal would not accurately reflect the contents of the data
> structure, so I'm not too thrilled with it.
Yeah -- looking at the script to generate node support functions[1], it
might be better go back to the original formulation (i.e., your proposed
patch), and then use a "path_hack4" for this struct member, which looks
similar to other hacks already there for other cases that require
bespoke handling.
[1] https://postgr.es/m/bee9fdb0-cd10-5fdb-3027-c4b5a240bc74@enterprisedb.com
--
Álvaro Herrera Breisgau, Deutschland — https://www.EnterpriseDB.com/