Re: generic options for explain - Mailing list pgsql-hackers

From Andres Freund
Subject Re: generic options for explain
Date
Msg-id 4A1B0538.20001@anarazel.de
Whole thread Raw
In response to Re: generic options for explain  (Tom Raney <twraney@comcast.net>)
List pgsql-hackers
Hi Tom,

On 05/25/2009 08:04 PM, Tom Raney wrote:
> So, why not put ALL interesting data in the EXPLAIN XML feed? I'm not
> suggesting for this discussion that we include discarded plans, but that
> we include every piece of data that may be of interest to folks building
> connecting tools. The parsers can pick and choose what they use easily
> and, because the feed isn't positional, won't break when addition data
> is added. A GUC parameter could govern the data included in this variant
> of EXPLAIN, but even that seems unnecessary. This approach will allow
> the standard EXPLAIN to evolve in whatever way pleases the humans
> without interfering with the machines.
Well, there is the problem Robert Haas described - some stats may be too 
expensive to gather (like the io-stats) for regular use, but still be 
quite usefull.

Andres


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: No sanity checking performed on binary TIME parameters.
Next
From: Robert Haas
Date:
Subject: Re: generic options for explain