Обсуждение: Error when using unquoted identifier that contains the word end in a begin atomic block
Error when using unquoted identifier that contains the word end in a begin atomic block
PostgreSQL 15.2 (Debian 15.2-1.pgdg110+1) on x86_64-pc-linux-gnu, compiled by gcc (Debian 10.2.1-6) 10.2.1 20210110, 64-bit
Run via Docker 15.4 image
It seems to me that within a BEGIN ATOMIC block the use of the word “end” within an unquoted identifier is causing an error.
The word “end” is a reserved keyword. However, it is my understanding from the documentation that keywords cannot be used *as* an identifier, but they should be able to be used within an identifier.
For instance, the “end” in effective_end_date is a valid identifier.
-- Create a data set with the word end in it.
CREATE VIEW test_view AS
SELECT CURRENT_TIMESTAMP as effective_end_date
> OK
> Query Time: 0.002s
-- This function is successfully created (note the quoted “effective_end_date”).
CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP
LANGUAGE SQL
BEGIN ATOMIC
SELECT DISTINCT "effective_end_date"
FROM test_view;
END
> Affected rows: 0
> Query Time: 0.001s
-- This function fails to be created. I would expect this to work.
CREATE OR REPLACE FUNCTION test_function_fail() RETURNS TIMESTAMP
LANGUAGE SQL
BEGIN ATOMIC
SELECT DISTINCT effective_end_date
FROM test_view
> ERROR: syntax error at end of input
LINE 6: FROM test_view
^
> Query Time: 0s
Jonathan Zacharuk <jonathan.zacharuk@ascentech.ca> writes:
> It seems to me that within a BEGIN ATOMIC block the use of the word "end" within an unquoted identifier is causing an
error.
I think your problem is somewhere on the client side.
> -- This function is successfully created (note the quoted "effective_end_date").
> CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP
> LANGUAGE SQL
> BEGIN ATOMIC
> SELECT DISTINCT "effective_end_date"
> FROM test_view;
> END
For me, this works (in psql) with or without the double quotes:
regression=# CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP
LANGUAGE SQL
BEGIN ATOMIC
SELECT DISTINCT "effective_end_date"
FROM test_view;
END;
CREATE FUNCTION
regression=# CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP
LANGUAGE SQL
BEGIN ATOMIC
SELECT DISTINCT effective_end_date
FROM test_view;
END;
CREATE FUNCTION
> -- This function fails to be created. I would expect this to work.
> CREATE OR REPLACE FUNCTION test_function_fail() RETURNS TIMESTAMP
> LANGUAGE SQL
> BEGIN ATOMIC
> SELECT DISTINCT effective_end_date
> FROM test_view
> ERROR: syntax error at end of input
> LINE 6: FROM test_view
> ^
I can't help noticing that the trailing END is missing from what you
show here, and it seems pretty clear that the syntax error is about
that, not about the effective_end_date identifier. So the real
question is why the CREATE FUNCTION command is getting sent to the
server before it's complete. If this isn't flat-out pilot error,
then I guess whatever client you are using (apparently not psql) is
getting confused about which occurrence of "end" terminates the
command. You should take this up with the authors of said
client software.
regards, tom lane
RE: Error when using unquoted identifier that contains the word end in a begin atomic block
Thanks Tom. I should have noticed that the command I was running was not completely getting sent to the server. This
doesindeed look like a Navicat issue. I have confirmed with psql that this works fine and will ensure to do so in the
future!
-----Original Message-----
From: Tom Lane <tgl@sss.pgh.pa.us>
Sent: Thursday, August 24, 2023 1:12 PM
To: Jonathan Zacharuk <jonathan.zacharuk@ascentech.ca>
Cc: pgsql-bugs@lists.postgresql.org
Subject: Re: Error when using unquoted identifier that contains the word end in a begin atomic block
Jonathan Zacharuk <jonathan.zacharuk@ascentech.ca> writes:
> It seems to me that within a BEGIN ATOMIC block the use of the word "end" within an unquoted identifier is causing an
error.
I think your problem is somewhere on the client side.
> -- This function is successfully created (note the quoted "effective_end_date").
> CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP
> LANGUAGE SQL BEGIN ATOMIC
> SELECT DISTINCT "effective_end_date"
> FROM test_view;
> END
For me, this works (in psql) with or without the double quotes:
regression=# CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP LANGUAGE SQL BEGIN ATOMIC
SELECT DISTINCT "effective_end_date"
FROM test_view;
END;
CREATE FUNCTION
regression=# CREATE OR REPLACE FUNCTION test_function_success() RETURNS TIMESTAMP LANGUAGE SQL BEGIN ATOMIC
SELECT DISTINCT effective_end_date
FROM test_view;
END;
CREATE FUNCTION
> -- This function fails to be created. I would expect this to work.
> CREATE OR REPLACE FUNCTION test_function_fail() RETURNS TIMESTAMP
> LANGUAGE SQL BEGIN ATOMIC
> SELECT DISTINCT effective_end_date
> FROM test_view
> ERROR: syntax error at end of input
> LINE 6: FROM test_view
> ^
I can't help noticing that the trailing END is missing from what you show here, and it seems pretty clear that the
syntaxerror is about that, not about the effective_end_date identifier. So the real question is why the CREATE
FUNCTIONcommand is getting sent to the server before it's complete. If this isn't flat-out pilot error, then I guess
whateverclient you are using (apparently not psql) is getting confused about which occurrence of "end" terminates the
command. You should take this up with the authors of said client software.
regards, tom lane