Hi,
On 2018-10-17 13:14:17 -0400, James Coleman wrote:
> > It's far from only toast that could be problematic here.
> >
>
> Do you have an example in mind? Wouldn’t the tuples have to be corrupted in
> some way to have problems with being interpreted as a datum? Or are you
> thinking very old tuples with a radically different structure to be causing
> the problem?
There's plenty ways it can go horribly wrong. Let's start with something
simple:
BEGIN;
ALTER TABLE ... ADD COLUMN blarg INT;
INSERT ... (blag) VALUES (132467890);
ROLLBACK;
ALTER TABLE ... ADD COLUMN blarg TEXT;
If you now read the table with your function you'll see a dead row that
will re-interpret a int datum as a text datum. Which in all likelyhood
will crash the server.
And yes, it's more likely to break with your patch, but I think the
current *split* function is dangerous too, and shouldn't have been added
without *A LOT* more attention to issues like this.
Greetings,
Andres Freund