[check_postgres] check_sequence broken in HEAD

[Glyn Astill]

----- Original Message -----
> From: Christoph Berg <christoph.berg at credativ.de>
> To: Glyn Astill <glynastill at yahoo.co.uk>
> Cc: "check_postgres at bucardo.org" <check_postgres at bucardo.org>
> Sent: Wednesday, 15 April 2015, 21:24
> Subject: Re: [check_postgres] check_sequence broken in HEAD
> 
> Re: Glyn Astill 2015-04-15 
> <2134242480.4340165.1429117023283.JavaMail.yahoo at mail.yahoo.com>
>>  Cheers, apologies about the no-ops; I made quite a hash of syncing up my 
> repo.
>> 
>>  I'm also seeing some issues with the newer fancy version 
> check_same_schema comparing a lot of oid and attnum based columns, as well as it 
> not taking into account that trigger names are not unique unless  their schema 
> & table name is also taken into account.  
> 
> I've had regression test problems with the checkpoint and cluster_id
> checks but these seemed unrelated to your and my fixes. What problem

> are you seeing?

Well mainly I'm seeing the comparison of items based on oid values instead of resolving them to some logical meaning or ignoring them.  An example is the function check comparing pg_proc.proargtypes, which doesn't necessarily match up for user created types (by that I mean not defined in pg_type.h).

Looks like that's mainly a simple case of adding to the exclude list, I'll take a look at adding what I think suits.


The other issue is the trigger check keying on trigger name without referencing the table name, so for example it flags up slony logtriggers because it compares a logtrigger on one table with a logtrigger with the same name on another.  I'll see if I can do something about that too.


Glyn