Beta 10 parser error for CREATE STATISTICS IF NOT EXISTS

On 2017/06/21 9:42, Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

Sample output:
psql (10beta1)
Type "help" for help.

o365logs=# select version();

version
--------------------------------------------------------------------------------

----------------------------
PostgreSQL 10beta1 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5
20150623
(Red Hat 4.8.5-11), 64-bit
(1 row)

o365logs=# CREATE STATISTICS IF NOT EXISTS logs_corrtest (dependencies) ON
record_type, operation FROM logs;
ERROR: syntax error at or near "NOT"
LINE 1: CREATE STATISTICS IF NOT EXISTS logs_corrtest (dependencies)...

Looks like a documentation bug if the authors of the feature actually
meant to implement the following syntax:

CREATE [ IF NOT EXISTS ] STATISTICS

create if not exists statistics words_stats on a, b from words;
CREATE STATISTICS

create if not exists statistics words_stats on a, b from words;
NOTICE: statistics object "words_stats" already exists, skipping
CREATE STATISTICS

If that's really what's intended, it seems a bit inconsistent with most
other commands and with DROP STATISTICS [ IF NOT EXISTS ] itself.

Thanks,
Amit

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs

On 2017/06/21 10:15, Amit Langote wrote:

On 2017/06/21 9:42, Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

Sample output:
psql (10beta1)
Type "help" for help.

o365logs=# select version();

version
--------------------------------------------------------------------------------

----------------------------
PostgreSQL 10beta1 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5
20150623
(Red Hat 4.8.5-11), 64-bit
(1 row)

o365logs=# CREATE STATISTICS IF NOT EXISTS logs_corrtest (dependencies) ON
record_type, operation FROM logs;
ERROR: syntax error at or near "NOT"
LINE 1: CREATE STATISTICS IF NOT EXISTS logs_corrtest (dependencies)...

Looks like a documentation bug if the authors of the feature actually
meant to implement the following syntax:

CREATE [ IF NOT EXISTS ] STATISTICS

create if not exists statistics words_stats on a, b from words;
CREATE STATISTICS

create if not exists statistics words_stats on a, b from words;
NOTICE: statistics object "words_stats" already exists, skipping
CREATE STATISTICS

If that's really what's intended, it seems a bit inconsistent with most
other commands and with DROP STATISTICS [ IF NOT EXISTS ] itself.

Here is a patch, just in case, that changes the grammar to accept the
following syntax instead of the current one:

CREATE STATISTICS [ IF NOT EXIST ] ...

Also added a test. Documentation already displays the above syntax, so no
update needed there.

Thanks,
Amit

Amit Langote <Langote_Amit_f8@lab.ntt.co.jp> writes:

On 2017/06/21 9:42, Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

Looks like a documentation bug if the authors of the feature actually
meant to implement the following syntax:
CREATE [ IF NOT EXISTS ] STATISTICS

Hm, that is what the grammar supports, but surely it's utterly
inconsistent with every other usage of IF NOT EXISTS. Even if
this was intended and the docs were not, we should fix the grammar
to match the docs not vice versa.

regards, tom lane

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs

Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

You're absolutely right, and this is a bug in the parser -- I probably
misplaced the IF NOT EXISTS clause while playing with Tomas' parser
changes. Here's the fix. I'm a bit troubled that this change doesn't
seem to affect any tests, so I'll add some before pushing.

(Now, "statistics" being plural would seem to call for CREATE STATISTICS
IF NOT EXIST, rather than EXISTS, but I'll put that thought aside on
account of it being just too weird ...)

--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On 2017/06/21 11:02, Alvaro Herrera wrote:

Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

You're absolutely right, and this is a bug in the parser -- I probably
misplaced the IF NOT EXISTS clause while playing with Tomas' parser
changes. Here's the fix. I'm a bit troubled that this change doesn't
seem to affect any tests, so I'll add some before pushing.

I posted the same patch but with a test:

/messages/by-id/38b7d52e-387a-0e47-7525-b8b654ca4bfb@lab.ntt.co.jp

Forgot to cc you or Tomas.

Thanks,
Amit

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs

Amit Langote wrote:

On 2017/06/21 11:02, Alvaro Herrera wrote:

Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

You're absolutely right, and this is a bug in the parser -- I probably
misplaced the IF NOT EXISTS clause while playing with Tomas' parser
changes. Here's the fix. I'm a bit troubled that this change doesn't
seem to affect any tests, so I'll add some before pushing.

I posted the same patch but with a test:

/messages/by-id/38b7d52e-387a-0e47-7525-b8b654ca4bfb@lab.ntt.co.jp

Pushed.

--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs

On 2017/06/23 2:22, Alvaro Herrera wrote:

Amit Langote wrote:

On 2017/06/21 11:02, Alvaro Herrera wrote:

Bruno Wolff III wrote:

I'm not seeing an obvious error in my attempt to use CREATE STATISTICS IF
NOT EXISTS. Given this is new, maybe there is a bug in the parser.

You're absolutely right, and this is a bug in the parser -- I probably
misplaced the IF NOT EXISTS clause while playing with Tomas' parser
changes. Here's the fix. I'm a bit troubled that this change doesn't
seem to affect any tests, so I'll add some before pushing.

I posted the same patch but with a test:

/messages/by-id/38b7d52e-387a-0e47-7525-b8b654ca4bfb@lab.ntt.co.jp

Pushed.

Thanks, Alvaro.

Regards,
Amit

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs