On 08.06.21 04:50, Julien Rouhaud wrote:
> On Mon, Jun 07, 2021 at 06:10:58PM -0400, Tom Lane wrote:
>>
>> I'm fairly disinclined to do anything about it though, because I'm
>> afraid of the SQL committee standardizing some other syntax for the
>> same idea in future (or maybe worse, commandeering the same keyword
>> for some other feature). It doesn't seem quite valuable enough to
>> take those risks for.
>
> Also, isn't the OP problem already solved by the SEARCH / CYCLE grammar
> handling added in 3696a600e2292?
You still get the path column in the output, which is what the OP didn't
want. But optionally eliminating the path column from the output might
be a more constrained problem to solve. We actually already discussed
this; we just need to do it somehow.