richard coleman <rcoleman.ascentgl@gmail.com> writes:
> Look below for the difference in the two functions, how they are used, the
> amount of knowledge you need to posses in order to run them and the
> difference in the output.
I'm not finding this terribly convincing, because I don't exactly see
what is the use-case for having that DDL in isolation. Or in other
words, what is your use-case that you find pg_dump so unsatisfactory
for? It knows all this stuff, and it also knows things like which
other objects a given object depends on and how to cope with circular
dependencies. Most of the reasons that I can think of for wanting
the pg_get_*def functions that we don't already have seem to reduce
to building partially-baked versions of pg_dump.
If your concern is that you can't get pg_dump to slice up the schema
in exactly the way you want, adding some more options to it might be
an easier sell than implementing/maintaining a bunch of duplicate
functionality. We just got done adding [1], for example.
regards, tom lane
[1] https://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=a563c24c9574b74f4883c004c89275bba03c3c26
