At 12PM -0800 on 20/02/13 you (Sergey Konoplev) wrote:
> On Wed, Feb 20, 2013 at 10:16 AM, Ben Morrow wrote:
> >> If you got mixed up with plpgsql anyway what is the reason of making
> >> this WITH query constructions instead of implementing everything in a
> >> plpgsql trigger on DELETE on exp th
On Wed, Feb 20, 2013 at 10:16 AM, Ben Morrow wrote:
>> If you got mixed up with plpgsql anyway what is the reason of making
>> this WITH query constructions instead of implementing everything in a
>> plpgsql trigger on DELETE on exp then?
>
> I'm not sure what you mean. "exp" isn't a table, it's a
At 8AM -0800 on 20/02/13 you (Sergey Konoplev) wrote:
> On Wed, Feb 20, 2013 at 12:19 AM, Ben Morrow wrote:
> > That's not reliable. A concurrent txn could insert a conflicting row
> > between the update and the insert, which would cause the insert to fail
> > with a unique constraint violation.
On Wed, Feb 20, 2013 at 12:19 AM, Ben Morrow wrote:
> That's not reliable. A concurrent txn could insert a conflicting row
> between the update and the insert, which would cause the insert to fail
> with a unique constraint violation.
Okay I think I got it. The function catches exception when INS
Quoth gray...@gmail.com (Sergey Konoplev):
> On Sat, Feb 16, 2013 at 11:58 PM, Ben Morrow wrote:
> > WITH "exp" AS ( -- as before
> > ),
> > "subst" AS (
> > SELECT add_item(e.basket, e.nref, e.count)
> > FROM "exp" e
> > WHERE e.nref IS NOT NULL
> > )
> >
