Horton

Thread

Commits

  1. Doc: clarify use of RECURSIVE in WITH.

  1. documentation for WITH RECURSIVE

    Andreas Kretschmer <andreas@a-kretschmer.de> — 2019-11-18T14:36:24Z

    Hi,

One of our customers had problems with the syntax of WITH RECURSIVE. He 
suggested that the documentation should be improved, maybe with this 
sentence:

"If any of the CTEs is recursive, the recursive keywork must be present 
after with", so to make sure that it does not refer to just one of the 
named subqueries.



Regards, Andreas

-- 
2ndQuadrant - The PostgreSQL Support Company.
www.2ndQuadrant.com

  2. Re: documentation for WITH RECURSIVE

    Tom Lane <tgl@sss.pgh.pa.us> — 2019-11-18T17:32:08Z

    Andreas Kretschmer <andreas@a-kretschmer.de> writes:
> One of our customers had problems with the syntax of WITH RECURSIVE. He 
> suggested that the documentation should be improved, maybe with this 
> sentence:
> "If any of the CTEs is recursive, the recursive keywork must be present 
> after with", so to make sure that it does not refer to just one of the 
> named subqueries.

Hm, was this intended for the SELECT man page (presumably in the
"WITH Clause" section), or for section 7.8?

In the SELECT page, I'd be inclined to put it after the para about
"queries need not be ordered" and phrase it more like

  If there are multiple queries in the WITH clause, RECURSIVE can
  be written only once, immediately after WITH.  It applies to all
  queries in the WITH clause (but has no effect on queries that do
  not use either of these features).

			regards, tom lane

  3. Re: documentation for WITH RECURSIVE

    Andreas Kretschmer <andreas@a-kretschmer.de> — 2019-11-18T19:12:09Z

    
Am 18.11.19 um 18:32 schrieb Tom Lane:
> Andreas Kretschmer <andreas@a-kretschmer.de> writes:
>> One of our customers had problems with the syntax of WITH RECURSIVE. He
>> suggested that the documentation should be improved, maybe with this
>> sentence:
>> "If any of the CTEs is recursive, the recursive keywork must be present
>> after with", so to make sure that it does not refer to just one of the
>> named subqueries.
> Hm, was this intended for the SELECT man page (presumably in the
> "WITH Clause" section), or for section 7.8?
>
> In the SELECT page, I'd be inclined to put it after the para about
> "queries need not be ordered" and phrase it more like
>
>    If there are multiple queries in the WITH clause, RECURSIVE can
>    be written only once, immediately after WITH.  It applies to all
>    queries in the WITH clause (but has no effect on queries that do
>    not use either of these features).
>
> 			regards, tom lane

yeah, sounds good to me, thx Tom.

Regards, Andreas

-- 
2ndQuadrant - The PostgreSQL Support Company.
www.2ndQuadrant.com

  4. Re: documentation for WITH RECURSIVE

    Andreas Kretschmer <andreas@a-kretschmer.de> — 2019-11-19T12:29:39Z

    
Am 18.11.19 um 20:12 schrieb Andreas Kretschmer:
>
>
> Am 18.11.19 um 18:32 schrieb Tom Lane:
>> Andreas Kretschmer <andreas@a-kretschmer.de> writes:
>>> One of our customers had problems with the syntax of WITH RECURSIVE. He
>>> suggested that the documentation should be improved, maybe with this
>>> sentence:
>>> "If any of the CTEs is recursive, the recursive keywork must be present
>>> after with", so to make sure that it does not refer to just one of the
>>> named subqueries.
>> Hm, was this intended for the SELECT man page (presumably in the
>> "WITH Clause" section), or for section 7.8?
>>
>

Hi Tom,

it would be good if we can also add this to chapter 7.8, wouldn't it?


Regards, Andreas

-- 
2ndQuadrant - The PostgreSQL Support Company.
www.2ndQuadrant.com

  5. Re: documentation for WITH RECURSIVE

    Tom Lane <tgl@sss.pgh.pa.us> — 2019-11-19T19:47:16Z

    Andreas Kretschmer <andreas@a-kretschmer.de> writes:
> it would be good if we can also add this to chapter 7.8, wouldn't it?

I don't feel a need to.  7.8 isn't attempting to define the syntax
precisely.  Moreover, we've had this feature since 8.4, and this
is the first I've heard of somebody misinterpreting the syntax;
so it doesn't seem like it's a huge problem that requires multiple
explanations.

I've pushed a patch for the SELECT reference page.

			regards, tom lane