I have the following query:
SELECT opstudy, product_set, RANK() OVER (PARTITION BY opstudy ORDER BY case_pack_qty) AS CASE_PK_RANK
FROM rngr_15_grp;
that I'm using in MS Access. I have the Object Designer settings set to SQL Server syntax compatible. I'm getting
"syntax error (missing operator) in query expression RANK() OVER (PARTITION BY opstudy ORDER BY case_pack_qty)
What operator belongs in that expression?
You can't use RANK() OVER ... in Access. The setting SQL Server Compatible Syntax (ANSI 92) implies that you must use ANSI-92 syntax for all queries. The RANK() analytic window function wasn't introduced in the ANSI standard until ANSI SQL:1999 and thus can't be used in SQL-92 mode.
Related
I'm using API 31 with Room v2.4.2 and SQLite v3.32.2 which is supposed to support window functions. But a simple query like this
select *, row_number() over(order by id) from mytable
produces the following error at over(
<compound operator>, FROM, GROUP, LIMIT, ORDER, WHERE, comma or semicolon expected, got '('
What's wrong with that?
The timescale docs seem to suggest that since 2.7.0 it should be possible to make materialized views which include an order by clause. (See "timescale.finalized" option here and "function support" here).
However, I have not been able to get this to work for me. When I try to create my materialized view I get:
ERROR: invalid continuous aggregate query
DETAIL: ORDER BY is not supported in queries defining continuous aggregates.
HINT: Use ORDER BY clauses in SELECTS from the continuous aggregate view instead.
Is there something fundamental I'm misunderstanding about how this should work?
Here is the full script:
> select extname, extversion from pg_extension where extname = 'timescaledb';
extname | extversion
-------------+------------
timescaledb | 2.7.0
(1 row)
> CREATE TABLE stocks_real_time (
time TIMESTAMPTZ NOT NULL,
price DOUBLE PRECISION NULL
);
CREATE TABLE
> SELECT create_hypertable('stocks_real_time','time');
create_hypertable
-------------------------------
(7,public,stocks_real_time,t)
(1 row)
> CREATE MATERIALIZED VIEW mat_view_stocks_real_time
WITH (timescaledb.continuous)
AS (
SELECT
time_bucket('60 minutes', time) as bucketed_time,
AVG(price) as price
FROM stocks_real_time
GROUP BY bucketed_time
ORDER BY bucketed_time
);
ERROR: invalid continuous aggregate query
DETAIL: ORDER BY is not supported in queries defining continuous aggregates.
HINT: Use ORDER BY clauses in SELECTS from the continuous aggregate view instead.
I still get the same error if I explicitly add "timescaledb.finalized=true" to the with clause.
(NB: I work at Timescale!)
We have an open issue to support this, and I think the confusion is because we now support aggregates with order by clauses in them, this means things like: SELECT percentile_cont(price) WITHIN GROUP (ORDER BY time) or SELECT array_agg(foo ORDER BY time)
So I think that is probably where the confusion is coming from, but like I said, we have an open issue to support that sort of order by. You can also apply the order by in the SELECT from the continuous aggregate though: ie SELECT * FROM mat_view_stocks_real_time ORDER BY bucketed_time and that should work just fine.
I want to allow sort to parameterized in a query, field id and order
so I have select .... order by #sortcol
then what? How to I use a parameter to indicate asc vs desc?
Using c# npgsql library.
I've got a table, me, that's got an e_id column, a first_name, and a last_name. (A few other columns too but I'm just trying to figure out the window function stuff.) When I try to do a query and pick out the first name values in the table for a given e_id value:
SELECT e_id, first_value(first_name), first_value(last_name)
OVER (PARTITION BY e_id)
FROM me
I get the "window function call requires an OVER clause" error. Now, like I said, I don't know what I'm doing yet, but I'm pretty sure there's at least an attempt at an OVER clause in that query. OK, so when I try it without the functions:
SELECT e_id, first_name, last_name
OVER (PARTITION BY e_id)
FROM me
I get a syntax error at OVER. I'm running psql version 9.4.4 against a 9.1.13 server. I'm staring at the documentation for 9.1 and it looks to me like OVER is documented there. Am I just missing something basic here?
Each window function must have its own OVER clause. The problem with your first query is that the first_value(first_name) window function does not have an OVER clause.
And the problem with your second query is that you have an OVER clause that is not preceded by a window function.
Try this
SELECT e_id,
first_value(first_name) OVER (PARTITION BY e_id),
first_value(last_name) OVER (PARTITION BY e_id)
FROM me
I'm fairly proficient in mySQL and MSSQL, but I'm just getting started with postgres. I'm sure this is a simple issue, so to be brief:
SQL error:
ERROR: column "incidents.open_date" must appear in the GROUP BY clause or be used in an aggregate function
In statement:
SELECT date(open_date), COUNT(*)
FROM incidents
GROUP BY 1
ORDER BY open_date
The type for open_date is timestamp with time zone, and I get the same results if I use GROUP BY date(open_date).
I've tried going over the postgres docs and some examples online, but everything seems to indicate that this should be valid.
The problem is with the unadorned open_date in the ORDER BY clause.
This should do it:
SELECT date(open_date), COUNT(*)
FROM incidents
GROUP BY date(open_date)
ORDER BY date(open_date);
This would also work (though I prefer not to use integers to refer to columns for maintenance reasons):
SELECT date(open_date), COUNT(*)
FROM incidents
GROUP BY 1
ORDER BY 1;
"open_date" is not in your select list, "date(open_date)" is.
Either of these will work:
order by date(open_date)
order by 1
You can also name your columns in the select statement, and then refer to that alias:
select date(open_date) "alias" ... order by alias
Some databases require the keyword, AS, before the alias in your select.