Re: Re: [INTERFACES] USE OF CURSORS IN ECPG - Mailing list pgsql-hackers

From Christof Petig
Subject Re: Re: [INTERFACES] USE OF CURSORS IN ECPG
Date
Msg-id 3A0BD380.4B5F37E@wtal.de
Whole thread Raw
In response to Re: [INTERFACES] USE OF CURSORS IN ECPG  ("Maurizio" <maurizio.c@libero.it>)
List pgsql-hackers
Maurizio wrote:

> But, how can I do ?
> I have to recall the same routine many times. I tried to prepare and declare
> the cursor inside the routine but when I compile with ecpg I receive the
> error Cursor already defined.

you should drop the cursor (exec sql close name;)
But there is trouble ahead (believe me - see below)

> Maurizioan

------------
(Oh, I misunderstood, if you get into trouble with transactions (waiting for
you ;-() you might consider the following lines)
------------
You certainly should disable autocommit!
  exec sql set autocommit to off;
otherwise you have no control over when a transaction starts and whether this
succeded.
Then at the start of your SubProgram do
  bool my_transaction;
  exec sql begin work;
  my_transaction=!sqlca.sqlcode;

at the end:
  if (my_transaction) exec sql commit work;

This should give you an idea on how to accomplish it. If not feel free to ask
me.

But (sigh) you need my NOTICE to sqlca patch applied. It's not in 7.0 but in
7.1! I don't know any other way (I'll attach the patch)

Christof

PS: You might consider applying the patch for (update where not_found) -> 100
and the cache which doubles ecpg's speed.


Attachment

pgsql-hackers by date:

Previous
From: Dirk Lutzebaeck
Date:
Subject: Must a C state transition function use palloc on the returned value?
Next
From: Zeugswetter Andreas SB
Date:
Subject: AW: Re: [INTERFACES] USE OF CURSORS IN ECPG