On Friday, September 19, 2003, at 12:21 AM, Tom Lane wrote:
> Andrew Rawnsley <ronz@ravensfield.com> writes:
>> Sequence information is stored in their individual 1-row tables, with
>> an entry is pg_class of relkind 'S'. So you can't
>> really get a list of sequences with last_value in a single query
>
> Nope, you can't ... and I surely hope you weren't expecting that the
> last_values would all be simultaneously valid ...
>
No, I don't expect that. I'm shooting for 'best I can do'.
> What I'd try for this is
>
> select relname, get_last_value(relname) from pg_class where relkind =
> 'S';
>
> where get_last_value() is a plpgsql function that does an EXECUTE.
> But you must realize that there will be time skew between the
> last_values.
>
> What is the problem you really want to solve?
>
Hacking some semblance of sequence support into eRserver. The possible
skew when gathering last_values
doesn't bother me too much - the replicated system is out of sync to
start with, and the hope is that everything will stay pretty
much caught up. Its certainly possible for me to be caught with my
pants down and have a sequence lagging the column its
supposed to represent at time of failure. But a) its not likely, and b)
I don't have much choice anyway. Something is better than
nothing. Pays your money you takes your chances.
I'll poke around with the function idea. Certain better than doing it
all from the client side...
> regards, tom lane
>
> ---------------------------(end of
> broadcast)---------------------------
> TIP 9: the planner will ignore your desire to choose an index scan if
> your
> joining column's datatypes do not match
>
--------------------
Andrew Rawnsley
President
The Ravensfield Digital Resource Group, Ltd.
(740) 587-0114
www.ravensfield.com
--------------------
Andrew Rawnsley
President
The Ravensfield Digital Resource Group, Ltd.
(740) 587-0114
www.ravensfield.com