On 7/21/2006 10:47 AM, Gavin Hamill wrote:
> Jan Wieck wrote:
>> I have attached a patch that I am about to commit into REL_1_1_STABLE 
>> as well as HEAD. It will cause slon_quote_literal() to return a NULL 
>> result for a NULL input and the log trigger will bail out with an 
>> ERROR if one of the old key columns contains the NULL value. This will 
>> cause a transaction that happens to touch such a row to abort, instead 
>> of causing a postgres core dump.
> 
> This is a followup to a thread back from mid-May [1] - I've been able to 
> apply your patch, and the great news is pg no longer segfaults - hurrah :)
> 
> What interests me is the output in the logs... When before pg would fall 
> over, now I just get
> 
> old key column User."DisplayNumberRooms" IS NULL on UPDATE
> 
> Now, I don't understand this at all... the User table has dozens of 
> columns, and DisplayNumberRooms is just a bool - not a primary or 
> foreign key; the column is not marked as NOT NULL so I can't understand 
> what the problem actually is here...

The problem is that the column is (most definitely) used in the index 
that you have specified for identifying the rows.


Jan

> 
> CREATE TABLE "User"
> (
> ....
>   "DisplayNumberRooms" bool DEFAULT false,
> ...
> )
> 
> Cheers,
> Gavin.
> 
> 
> [1] 
> http://gborg.postgresql.org/pipermail/slony1-general/2006-May/004272.html
> _______________________________________________
> Slony1-general mailing list
> Slony1-general at gborg.postgresql.org
> http://gborg.postgresql.org/mailman/listinfo/slony1-general


-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #