On 08/20/2014 07:30 PM, Tom Lane wrote:
> I wrote:
>> I think returning NULL is the right thing here, really. Aside from being
>> arguably more convenient for indexing, we will get less push-back if we
>> make some operators go from throwing errors to returning null than if
>> we make some other operators go the other way.
> Attached is a proposed patch (against HEAD and 9.4 git tip) that has
> these effects:
>
> 1. #> now returns the input object if the RHS array is empty;
So will
val #>> '{}'
now return a dequoted bare scalar string? I think that's where the OP
actually came into this.
>
> 2. Error cases in -> and #> are all removed in favor of returning NULL.
>
> The regression test changes show that we were really pretty inconsistent
> about whether to throw error or return NULL for cases where the JSON
> tree structure didn't match the extraction request. I think having a
> uniform rule is a definite improvement.
>
>
Works for me. Thans for cleaning this up.
cheers
andrew