Re: Reducing some DDL Locks to ShareLock

Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
> On Tue, 2008-10-07 at 11:15 -0400, Tom Lane wrote:
>> Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
>>> 2. Also need to decide whether we want pg_class.reltriggers as int2 (as
>>> implemented here) or switch to relhastriggers as boolean.
>>
>> I'd go for changing the column name/type. Yeah, you will break any
>> clients that are still trying to manipulate reltriggers directly, but
>> better to break them obviously than non-obviously. And I think a silent
>> change in the column semantics has significant risk of the latter.

> New version with column type change.

I'm starting to review this now. It strikes me that while we are at it,
we should get rid of the useless pg_class columns relukeys, relfkeys,
and relrefs. These haven't been maintained since Berkeley days, and
this patch puts the final kibosh on any thought that we'd ever start
to maintain relukeys and relfkeys counts.

Any objections?

regards, tom lane