On Sun, Nov 15, 2015 at 1:34 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Michael Paquier <michael.paquier@gmail.com> writes:
>> On Sun, Nov 1, 2015 at 2:03 AM, Julien Rouhaud
>> <julien.rouhaud@dalibo.com> wrote:
>>> I'm also rather sceptical about this change.
>
>> Hm. Thinking a bit about this patch, it presents the advantage to be
>> able to track the same queries easily across systems even if those
>> tables have been created with a different OID.
>
> That argument would only hold if *every* use of OIDs in the jumbles
> were replaced by names --- not only tables, but types, operators,
> functions, etc. I'm already concerned that the additional name
> lookups will cost a lot of performance, and I think adding so many
> more would probably be disastrous.
Yeah, I was thinking about a GUC switch to change from one mode to
another yesterday night before sleeping. Now if there was a patch
actually implementing that, and proving that this has no performance
impact, well I think that this may be worth considering for
integration. But we're far from that
>> Also, isn't this patch actually broken if we rename relations and/or
>> its namespace?
>
> Well, it would mean that the query would no longer be considered "the
> same". You could argue either way whether that's good or bad. But
> yeah, this approach will break one set of use-cases to enable another
> set.
>
> On the whole, I think my vote is to reject this patch.
Agreed. It's clear that the patch as-is is not enough.
--
Michael
