Обсуждение: pgvector extension error

"Zacher, Stacy" <szacher@mcw.edu> writes:
> We are having issues getting the pgvector extension to work in a database.

This:

> dbdev=# create extension if not exists vector;
> ERROR:  type "vector" already exists

indicates you've got some type named "vector" in that database
already.  It didn't come from pgvector, so where did it come from?

            regards, tom lane

I want to concatenate the hstores,

On 1/17/25 12:43, Brent Wood wrote:
> Hi,
> 
> I have a table with a timestamp and hstore columns with times in 
> milliseconds.
> 
> I want to concatenate the hstores, grouped by timestamp, with the 
> timestamps truncated to whole seconds: date_trunc('seconds', timer).
> 
> How can I concatenate all the hstores within a one second interval,
> 
>   ie, something like:
> 
> select timer,
>                <hstores concatenated>
> from (select  date_trunc('seconds', timer) as timer_sec, hstore)
> group by timer;

1) This is going to need some example data and the table definition.

2) The above query does not make sense.

3) Hstore does not store duplicate key values:

 From here:

https://www.postgresql.org/docs/current/hstore.html#HSTORE-OPS-FUNCS

select 'a=>b, c=>d'::hstore || 'c=>x, d=>q'::hstore ;
            ?column?
------------------------------
  "a"=>"b", "c"=>"x", "d"=>"q"

Are there duplicated keys and if so how do you want to handle them?


> 
> 
> Thanks
> 
> 
> Brent Wood
> 
> Principal Technician, Fisheries
> NIWA
> DDI:  +64 (4) 3860529
> 


-- 
Adrian Klaver
adrian.klaver@aklaver.com

"David G. Johnston" <david.g.johnston@gmail.com> writes:
> On Friday, January 17, 2025, Brent Wood <Brent.Wood@niwa.co.nz> wrote:
>> I want to concatenate the hstores,

> There are no hstore aggregate functions.

It would be trivial to make a user-defined one.
More or less (untested)

create aggregate hstore_agg(hstore)
(
  sfunc = hs_concat,
  stype = hstore
);

Either way though, the order of aggregation is unspecified and thus
it's unclear which input will win when there are duplicate keys.
If that matters then you'll need to think about how you want
to resolve it.

            regards, tom lane

On 1/19/25 12:09, Brent Wood wrote:
> Thanks for the replies, appreciated...
> 
> My current solution is:
> 
> /select trip_code,/
> /            station_no,/
> /            timer_sec + interval '12 hour' as NZST,/
> /            timer_sec as utc,/
> /            hstore_to_json(string_agg(values_sec::text, ', ')::hstore) 
> as values_sec/
> /     from (select '$TRIP' as trip_code,/
> /                  $STATION as station_no,/
> /                  date_trunc('second', timer) as timer_sec,/
> /                  values_sec/
> /           from t_reading_hstore_sec/
> /           where timer >= '$ISO_S'::timestamp - interval '12 hour'/
> /             and timer <= '$ISO_F'::timestamp - interval '12 hour') as foo/
> /group by timer_sec, trip_code, station_no;/
> 
> Convert the hstore to text, aggregate the text with string_agg(), 
> convert back to hstore (which seems to remove duplicate keys, OK for my 
> purpose)

To be clear values_sec in t_reading_hstore_sec is the hstore field?

If so what is it's structure?

> and group by timer truncated to whole seconds. I also provide UTC & 
> local timezone times for each set of readings. It is run in a bash 
> script which passes the trip & station values to the query, as well as 
> the start/finish times as ISO format strings.
> 
> The output is going to a Sqlite3 (Spatialite) database, which does not 
> have hstore, or all the hstore functionality that Postgres has, but does 
> have a json datatype which is adequate for our purposes, hence the 
> hstore_to_json in the query.
> 
> 
> Thanks again,
> 
> Brent Wood
> 

-- 
Adrian Klaver
adrian.klaver@aklaver.com

On 1/21/25 10:02, Brent Wood wrote:
> Yes, a Timescale hypertable with 500,000,000 rows of about 15 key/value 
> pairs per record.
> 
> I'm not sure why there is both a gin & gist index on the hstore, or the 
> merits of each.

Questions:

1) What is the timezone setting for the database?

2) What is the system time zone where the BASH script is run?

3) Are you sure you don't need the duplicate keys?
The keys maybe duplicated, but are the values duplicated also?

Also could this be simplified to something like?:

create table hstore_tbl (id integer, tsz_fld timestamptz, hstore_fld 
hstore);

insert into hstore_tbl values (1, now(), 'a=>1, b=>2'), (2, now() + 
interval '0.5 sec', 'a=>3, b=>4'), (3, now() + interval '1 sec', 'a=>1, 
b=>5'), (4, now()
+ interval '1.5 sec', 'a=>6, b=>7');

select id, date_trunc('second', tsz_fld) AS t_sec, h.key, h.value from 
hstore_tbl,  each(hstore_fld) as h ;
  id |         t_sec          | key | value
----+------------------------+-----+-------
   1 | 2025-01-21 14:00:27-08 | a   | 1
   1 | 2025-01-21 14:00:27-08 | b   | 2
   2 | 2025-01-21 14:00:28-08 | a   | 3
   2 | 2025-01-21 14:00:28-08 | b   | 4
   3 | 2025-01-21 14:00:28-08 | a   | 1
   3 | 2025-01-21 14:00:28-08 | b   | 5
   4 | 2025-01-21 14:00:29-08 | a   | 6
   4 | 2025-01-21 14:00:29-08 | b   | 7

It would unnest the data.

> 
> 
> Thanks....
> 
>   \d t_reading_hstore_sec
>                                           Table 
> "public.t_reading_hstore_sec"
>     Column   |            Type             | Collation | Nullable 
> |                      Default
>
------------+-----------------------------+-----------+----------+---------------------------------------------------
>   key        | bigint                      |           | not null | 
> nextval('t_reading_hstore_sec_key_seq'::regclass)
>   timer      | timestamp without time zone |           | not null |
>   values_sec | hstore                      |           |          |
> Indexes:
>      "t_reading_hstore_sec_pkey" PRIMARY KEY, btree (key, timer)
>      "t_reading_hstore_sec_timer_idx" btree (timer)
>      "t_reading_hstore_sec_timer_key" UNIQUE CONSTRAINT, btree (timer)
>      "t_reading_hstore_sec_values_idx_gin" gin (values_sec)
>      "t_reading_hstore_sec_values_idx_gist" gist (values_sec)
> 
> 
> 
> 
> On Wednesday, January 22, 2025 at 06:34:38 AM GMT+13, Adrian Klaver 
> <adrian.klaver@aklaver.com> wrote:
> 
> 
> On 1/19/25 12:09, Brent Wood wrote:
>  > Thanks for the replies, appreciated...
>  >
>  > My current solution is:
>  >
>  > /select trip_code,/
>  > /            station_no,/
>  > /            timer_sec + interval '12 hour' as NZST,/
>  > /            timer_sec as utc,/
>  > /            hstore_to_json(string_agg(values_sec::text, ', ')::hstore)
>  > as values_sec/
>  > /     from (select '$TRIP' as trip_code,/
>  > /                  $STATION as station_no,/
>  > /                  date_trunc('second', timer) as timer_sec,/
>  > /                  values_sec/
>  > /           from t_reading_hstore_sec/
>  > /           where timer >= '$ISO_S'::timestamp - interval '12 hour'/
>  > /             and timer <= '$ISO_F'::timestamp - interval '12 hour') 
> as foo/
>  > /group by timer_sec, trip_code, station_no;/
>  >
>  > Convert the hstore to text, aggregate the text with string_agg(),
>  > convert back to hstore (which seems to remove duplicate keys, OK for my
>  > purpose)
> 
> To be clear values_sec in t_reading_hstore_sec is the hstore field?
> 
> If so what is it's structure?
> 
> 
>  > and group by timer truncated to whole seconds. I also provide UTC &
>  > local timezone times for each set of readings. It is run in a bash
>  > script which passes the trip & station values to the query, as well as
>  > the start/finish times as ISO format strings.
>  >
>  > The output is going to a Sqlite3 (Spatialite) database, which does not
>  > have hstore, or all the hstore functionality that Postgres has, but does
>  > have a json datatype which is adequate for our purposes, hence the
>  > hstore_to_json in the query.
>  >
>  >
>  > Thanks again,
>  >
>  > Brent Wood
> 
>  >
> 
> -- 
> Adrian Klaver
> adrian.klaver@aklaver.com <mailto:adrian.klaver@aklaver.com>
> 
> 
> 
> 

-- 
Adrian Klaver
adrian.klaver@aklaver.com