Re: pg_restore depending on user functions
Tom Lane <tgl@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Дмитрий Иванов <firstdismay@gmail.com>
Cc: pgsql-general@lists.postgresql.org
Date: 2021-11-15T00:06:11Z
Lists: pgsql-bugs, pgsql-general
=?UTF-8?B?0JTQvNC40YLRgNC40Lkg0JjQstCw0L3QvtCy?= <firstdismay@gmail.com> writes: > вс, 14 нояб. 2021 г. в 23:46, Tom Lane <tgl@sss.pgh.pa.us>: >> This is not "details", this is an evidence-free assertion. Please show >> a concrete example of problematic functions. > Ok, I'll do it. > [ example ] This is pretty far from being a self-contained example; I had to guess at the definitions of several types that you didn't provide. Having done so, though, the set of functions and types seem to dump and restore just fine. So I still don't see any problem here. Possibly worth noting is that pg_dump is careful to include this in its output script: SET check_function_bodies = false; Without that, I can believe that you'd have some trouble restoring these functions, since pg_dump doesn't know anything about the cross-references appearing in the function bodies. But with that, there don't appear to be any troublesome cross-references here. regards, tom lane
Commits
-
Adjust pg_dump's priority ordering for casts.
- d4f6a36d8265 13.6 landed
- b55f2b692655 15.0 landed
- 69949ea68cf9 12.10 landed
- 54619a25df7c 11.15 landed
- 2b36973d95c6 10.20 landed
- 0fdf67476ccb 14.2 landed