Hello Peter,
A possible option 3 is to keep the function in pgcrypto but change
its C code to call the core code.
Updated patch with this change included.
Patch applies cleanly, compiles (both pg and pgcrypto). make check (global
and pgcrypto) ok. Doc generation ok. Minor comments:
About doc: I'd consider "generation" instead of "generating" as a
secondary index term.
(There is also precedent for redirecting the extension function to the
internal one by changing the SQL-level function definition using CREATE
OR REPLACE FUNCTION ... LANGUAGE INTERNAL. But that seems more
complicated and would require a new extension version. It could maybe
be included if the extension version is changed for other reasons.)
What about avoiding a redirection with something like:
Datum (* const pg_random_uuid)(PG_FUNCTION_ARGS) = gen_random_uuid;
--
Fabien.
