On 02/26/2015 10:38 AM, Tom Lane wrote:
>
> Yeah, I think so. The sequence 'infinity'::timestamp to JSON to
> ISO-8601-only consumer is going to fail no matter what; there is no
> need for us to force a failure at the first step. Especially when
> doing so excludes other, perfectly useful use-cases.
>
> So +1 for removing the error and emitting "infinity" suitably quoted.
> Andrew, will you do that?
>
>
Yeah.
cheers
andrew