Horton

Re: [HACKERS] Cache lookup errors with functions manipulation object addresses

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: Michael Paquier <michael.paquier@gmail.com>, Thomas Munro <thomas.munro@enterprisedb.com>, Aleksander Alekseev <a.alekseev@postgrespro.ru>, PostgreSQL mailing lists <pgsql-hackers@postgresql.org>
Date: 2018-02-19T00:39:50Z
Lists: pgsql-hackers 
On Sat, Feb 17, 2018 at 07:17:24PM -0300, Alvaro Herrera wrote:
> Pushed 0003.

Thanks.

> Maybe we can get rid of format_type_be_qualified too, but I didn't
> care too much about it either; it's not a big deal I think. 

Agreed. There are 16 callers spread in objectaddress.c and regproc.c,
this would generate some diffs.  If there are extra opinions later on,
we could always revisit that.  The new API is modular enough anyway.

> What interested me more was whether we could get rid of the
> FORMAT_TYPE_TYPEMOD_GIVEN flag, but ended up deciding not to pursue that
> as a phenomenal waste of time.  Here are some references in case you
> care.
> 
> https://www.postgresql.org/message-id/flat/200111101659.fAAGxKX06044%40postgresql.org
> https://git.postgresql.org/pg/commitdiff/a585c20d12d0e22befc8308e9f8ccb6f54a5df69

Thanks for the threads, I didn't know about them.  I thought as well
about trying to remove FORMAT_TYPE_TYPEMOD_GIVEN but avoided to do so,
so as not to break things the way they should be for a long time as this
code is as it is now for at least as long as I am working on Postgres.
I didn't check the git history to see the logic behind the code though,
which I really should have done.  So thanks for the references.
--
Michael

Commits

  1. Eliminate cache lookup errors in SQL functions for object addresses

  2. Refactor routines for name lookups of procedures and operators

  3. Add new flag to format_type_extended() to get NULL for undefined type

  4. Introduce new extended routines for FDW and foreign server lookups

  5. Refactor routines for subscription and publication lookups

  6. get_relid_attribute_name is dead, long live get_attname

  7. Tweak parser so that there is a defined representation for datatypes