Re: [PATCH] bugfix for int2vectorin

Tom,
Thanks for the comments.

Caleb Welton <cwelton(at)greenplum(dot)com> writes:
> I believe the int2vectorin function should handle invalid input more cleanly.

On 12/1/09 7:38 PM, "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Why bother?

Because correctness is good. Results that produce unexpected results from incorrect input formatting lead to incorrect data being loaded, which is bad.

>> Under what circumstances would users (or anyone at all) be putting data into an int2vector?

I agree in principle, but users do the darnedest things. Perhaps they simply want a compact int array that won't toast. Perhaps they liked that word "vector" in the name. I stopped trying to second guess them a long time ago.

>> It's an internal type and is certainly not meant for use in user tables.

But there is nothing preventing it from being used in user tables so calling it "internal" doesn't really mean a whole lot.

What exactly is your objection to having the int2arrayin parser handle its input conversion reasonably?

-Caleb