|From:||Michael Paquier <michael(at)paquier(dot)xyz>|
|To:||Tomas Vondra <tomas(dot)vondra(at)2ndquadrant(dot)com>|
|Cc:||Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>|
|Subject:||Re: remove pg_class.relhaspkey|
|Views:||Raw Message | Whole Thread | Download mbox | Resend email|
On Sat, Mar 10, 2018 at 01:52:56PM +0100, Tomas Vondra wrote:
> I agree with this sentiment - I don't think those flags are particularly
> helpful for client applications, and would vote +1 for removal.
OK, so I can see that we are moving to a consensus here.
> For the other flags we would probably need to test what impact would it
> have (e.g. table with no indexes, many indexes on other tables, and
> something calling get_relation_info a lot). But this patch proposes to
> remove only relhaspkey.
Yes, you are right here. Peter, would you do that within this commit
fest or not? As we are half-way through the commit fest, we could also
cut the apple in half and just remove relhaspkey for now as that's a
no-brainer. So I would suggest to just do the latter and consider this
patch as done.
Attached is a rebased patch, there were some conflicts in pg_class.h by
|Next Message||Amit Langote||2018-03-14 05:54:57||Re: inserts into partitioned table may cause crash|
|Previous Message||Pavan Deolasee||2018-03-14 05:44:24||Re: Faster inserts with mostly-monotonically increasing values|