Hello,First let me congratulate everyone on Postgres 7. It is (functionality
wise)
MUCH richer than Sybase 12. I am in the midst of porting across an old
(well. oldish)
web database application to Postgres from Sybase, and in everything apart
from Speed
Postgres wins (then again, that is probably why Sybase wins out. It doesnt
have any
'nice' features. gah!).
the problem arises when I tried to translate a relatively simple trigger
from
Sybase to Postgres. The Sybase code looks like this :
create trigger upd_client_hist on client for insert
asdeclare @historyid integerdeclare @inserted_client_id varchar(16)declare @inserted_date_id datetime
select @inserted_client_id=CLIENT_ID from insertedselect @inserted_date_id=CHANGE_DATE from insertedselect
@historyid=max(history_id)+1from client where CLIENT_ID=@inserted_client_id update client set
history_id=@historyidwhereCLIENT_ID=@inserted_client_idand CHANGE_DATE=@inserted_date_id
go
(the go at the end jst executes the creation process).
I looked through all the documentation on Postgres and decided to make it
into a function. However, since this is a function I wanted to make the
function
'generic' to a degree. so i tried this:
CREATE FUNCTION upd_history_no_location(text,varchar(16))
RETURNS integer
AS 'SELECT max(history_id)+1 FROM $1 where CLIENT_ID=CAST($2 as varchar(16));'
LANGUAGE 'sql';
and that jst plain doesnt compile. i have tried and installed the plpgsql
language
as well on the database, but it doesnt seem to help any either. I plan on
tying
this into a trigger on INSERT (there is no update allowed on the table).
I cant use an autoincremental field, as i need each new 'client_id' to start
off
with a history of 1, and keep its own 'unique history field'.
its a historical database tied into a client workorder schema.
Any help ?
does this make sense even ;D
many thanks,
Stefs
