Re: ROWTYPE initialization question - Mailing list pgsql-general

From Alban Hertroys
Subject Re: ROWTYPE initialization question
Date
Msg-id 455AD8C6.5000801@magproductions.nl
Whole thread Raw
In response to Re: ROWTYPE initialization question  ("Jim C. Nasby" <jim@nasby.net>)
Responses Re: ROWTYPE initialization question  (Jim Nasby <jim@nasby.net>)
List pgsql-general
Jim C. Nasby wrote:
> On Thu, Nov 09, 2006 at 04:37:23PM +0100, Alban Hertroys wrote:
>> 'lo list,
>>
>> I have a plpgsql SP where I loop through a cursor. I have an internal
>> variable that keeps the previous row, so that I can compare it with the
>> current row in the cursor.
>> Like so;
>>
>> DECLARE
>>     current table%ROWTYPE;
>>     previous table%ROWTYPE;
>> BEGIN
>>     LOOP
>>     FETCH tableCur INTO current;
>>
>>     -- Do stuff
>>
>> But, in this loop I need to compare the previous row to the current one.
>> To do that I need to know whether a row was assigned to 'previous', or
>> there'll be very little to compare (if it doesn't throw an error).
>> How do I check for that?
>>
>> I guess it would be like this, but I'd like to be sure.
>>
>>     IF previous IS NOT NULL
>>     THEN
>>         -- Compare previous and current column values
>>     END IF
>>
>>     previous := current;
>>     END LOOP;
>> END;
>
> Try it and see. :) I think that will work.
>
> Might be better to just capture the error.

Reality caught up, I had to put this procedure in our development
database or people wouldn't be able to continue their work. It seems to
work.

I suppose the real question is this: As it is not possible to initialize
a %ROWTYPE type variable to NULL, is comparing it to NULL valid or is it
comparing apples and oranges? Does it yield the expected result (true if
the %ROWTYPE variable is undefined, false once it is defined)?

It seems to work as is, but this part of PL/PgSQL seems to be a bit
unspecific. It might as well have worked by using the FOUND special
variable, or have yielded an error (which it didn't).

Looks like I'll need to device some test cases to prove how the various
subtleties behave. Now where did I put that can of time...

--
Alban Hertroys
alban@magproductions.nl

magproductions b.v.

T: ++31(0)534346874
F: ++31(0)534346876
M:
I: www.magproductions.nl
A: Postbus 416
   7500 AK Enschede

// Integrate Your World //

pgsql-general by date:

Previous
From: Devrim GUNDUZ
Date:
Subject: Re: empty folder for downlaoding PostgreSQL 8.1.5 for FC
Next
From: Joe Conway
Date:
Subject: Re: connectby usage question