[postgis-users] Vacuum analyse ERROR: Operation ontwoGEOMETRIESwith different SRIDs

[Mark Cave-Ayland]

> -----Original Message-----
> From: postgis-users-bounces at postgis.refractions.net 
> [mailto:postgis-users-bounces at postgis.refractions.net] On 
> Behalf Of strk at refractions.net
> Sent: 15 June 2005 12:47
> To: 'PostGIS Users Discussion'
> Subject: Re: [postgis-users] Vacuum analyse ERROR: Operation 
> ontwoGEOMETRIESwith different SRIDs
> 
> 
> Mark, I think your one is a good workaround, nonetheless 
> having analyze fail on full database is not very kind.
> 
> Does anyone intentionally use btree operators with postgis ?
> If they are really just for system use we can rethink their 
> semantic, make it quiet or issue a NOTICE.
> 
> --strk;


Hi strk,

Yeah, in this case I think the application should be able to work around it,
but that's perhaps a different argument ;) I agree that clobbering users in
an ANALYZE is not a friendly thing to do, as if tables don't get ANALYZEd
then your system can end up running quite badly - but then again this is the
first complaint we've had about this so far.

As a compromise, how about just changing the semantics for the btree "less
than" and "equals" operators to the following:

	i) PG 7.1 - issue a NOTICE but continue
	ii) PG > 7.1 and PG < 8.0 - issue a WARNING but continue

As long as we document this in the relevant sections of the btree ops code
then we should be OK, since these versions are a fixed target and so we can
guarantee that no other operators will get used as part of the ANALYZE. I
don't think I would like to lessen the strictness checks any less than this
though, and it's probably not good to start shouting in the docs that
mixed-SRID columns are a supported feature ;)


Kind regards,

Mark.

------------------------
WebBased Ltd
17 Research Way
Tamar Science Park
Plymouth
PL6 8BT 

T: +44 (0)1752 797131
F: +44 (0)1752 791023
W: http://www.webbased.co.uk