Re: Please recommend postgresql.conf improvements for osm2pgsql loading Europe

Thank you for the insights

Re: Please recommend postgresql.conf improvements for osm2pgsql loading Europe

Turning fsync = off has resulted in no noticable build time reduction for my Dockerfile with OSM Europe data, but still thank you for the suggestion! >

Re: Please recommend postgresql.conf improvements for osm2pgsql loading Europe

Thank you, Justin - On Sat, Mar 30, 2024 at 4:33 AM Justin Clift wrote: > On 2024-03-30 05:53, Alexander Farber wrote: > > I use the following postgresql.conf in my Dockerfile > > ( the full version at https://stackoverflow.com/a/78243530/165071 ), > > when loading a

Please recommend postgresql.conf improvements for osm2pgsql loading Europe

Good evening, I use the following postgresql.conf in my Dockerfile ( the full version at https://stackoverflow.com/a/78243530/165071 ), when loading a 28 GByte large europe-latest.osm.pbf into PostgreSQL 16 with PostGIS extension: echo "shared_buffers = 1GB">>

Should I add a new schema for my application?

Good morning, I am seeking for a general advice if it would be a good idea for a PostgreSQL/PostGIS using application to add a third schema. I am using postgis/postgis Docker image which comes with "public" schema. Then I have loaded OpenStreetMap data into the database and have created

Re: Displaying chat by punished users only to themselves (db fiddle attached)

Thank you, that was it!

Re: Displaying chat by punished users only to themselves (db fiddle attached)

Good evening, I still have a problem with my JOIN expression - when I add more games, then messages from other games are displayed: https://dbfiddle.uk/?rdbms=postgres_14=e2ff211f59090d1eeab879498148f907 CREATE OR REPLACE FUNCTION words_get_chat( in_gid integer,

Re: Displaying chat by punished users only to themselves (db fiddle attached)

Good morning, this is a very insightful comment (among many) by you, David - On Wed, May 4, 2022 at 5:40 PM David G. Johnston wrote: > Well, that is basically why I was going on about the oddity of having > social be a part of the main query. Personally I would write it as > "myself.uid =

Re: Displaying chat by punished users only to themselves (db fiddle attached)

David, thanks but what do you mean by the last comment - On Wed, May 4, 2022 at 7:44 PM David G. Johnston wrote: > Using (SELECT uid FROM myself) provides the same result without the > from/join reference; the usage in the case and the where clause could be > rewritten to use opponent.uid so

Re: Displaying chat by punished users only to themselves (db fiddle attached)

Is that the right way to do it? https://dbfiddle.uk/?rdbms=postgres_14=7bd74243397da61ddc4c216ad919c7cc WITH myself AS ( SELECT uid FROM words_social WHERE social = in_social AND sid = in_sid LIMIT 1 ) SELECT

Re: Displaying chat by punished users only to themselves (db fiddle attached)

I think I am very close with the following CTE, but do not understand how to bring it into the main SELECT query: https://dbfiddle.uk/?rdbms=postgres_14=ee264dc98b44dee75aa4523164a327b3 WITH myself AS ( SELECT uid FROM words_social WHERE social =

Re: Displaying chat by punished users only to themselves (db fiddle attached)

I have tried CROSS JOIN and CASE WHEN (why be greedy, right?): https://dbfiddle.uk/?rdbms=postgres_14=43a33374d15a9330145007702138822c WITH myself AS ( SELECT uid FROM words_social WHERE social = in_social AND sid = in_sid LIMIT

Re: Displaying chat by punished users only to themselves (db fiddle attached)

I try with a CTE but cannot figure the syntax: https://dbfiddle.uk/?rdbms=postgres_14=acd6d06a7ea2efc73a0771530832d77e WITH cte AS ( SELECT uid FROM words_social WHERE social = in_social AND sid = in_sid LIMIT 1 )

Re: Displaying chat by punished users only to themselves (db fiddle attached)

I am probably needing LEFT JOIN LATERAL here (and am completely lost)? Or to switch to CTE as you suggest

Re: Displaying chat by punished users only to themselves (db fiddle attached)

David, I try then the following - On Wed, May 4, 2022 at 5:28 PM David G. Johnston wrote: > You missed quoting the part where I describe the on clauses you need to > distinguish between "them" and "me" > > Me: u.uid in (player...) and (s.uid = u.uid) > Them: u.uid in (player...) and (s.uid <>

Re: Displaying chat by punished users only to themselves (db fiddle attached)

David, I am trying your suggestion: On Wed, May 4, 2022 at 4:27 PM David G. Johnston wrote: > Assuming the base query is capable of returning all related chat messages > for both users (I'd probably place that portion into a CTE) the rows you > want to filter out are those whose c.uid is not my

Re: Displaying chat by punished users only to themselves (db fiddle attached)

My real SQL function has one more param, an "auth" string generated by my game, which complements the social network id "sid". I have just omitted it in my test case. >

Re: Displaying chat by punished users only to themselves (db fiddle attached)

Hi Ron, On Wed, May 4, 2022 at 4:56 PM Ron wrote: > > How do other web sites know to present only "my" data, even though they > don't encode "my" user id in the URL? > > that is the usual pattern with OAuth provided by: Facebook, Google, Amazon, Huawei, etc... After you auth with them in a

Re: Displaying chat by punished users only to themselves (db fiddle attached)

Thank you for replying, David! The "social dynamic" is needed, because I cannot pass real user id (via HTTP) to SQL queries. Instead I pass social network type "social" (like 100 is facebook, 200 is twitter) and the social network id "sid" returned by that network. This way noone can read chats

Displaying chat by punished users only to themselves (db fiddle attached)

Hello, I have developed a complete SQL fiddle for my question: https://dbfiddle.uk/?rdbms=postgres_14=dcf063ba1615b392cc3cfa347a32c97b The context is that I run an online game for two players using a PostgreSQL 14.2 backend. I would like to make my game more friendly by hiding chat messages of

Re: if not exists (SELECT 1... UNION SELECT 1...)

s/grumbling/wondering/

Re: if not exists (SELECT 1... UNION SELECT 1...)

Ah, thank you Ralf! That has explained it (because I was still grumbling...) On Wed, Feb 16, 2022 at 4:44 PM Ralf Schuchardt wrote: > You must rather end the assignment with a semicolon: > > out_text := '___WRONG___'; > RETURN; > > Otherwise what really happens is: > > out_text := '___WRONG___'

Re: if not exists (SELECT 1... UNION SELECT 1...)

Ah, I have to do RETURN '___WRONG___'; and not out_text := '___WRONG___' RETURN;

Re: if not exists (SELECT 1... UNION SELECT 1...)

Hi Laurenz, thanks for your reply, but I think it is wrong - On Wed, Feb 16, 2022 at 1:24 PM Laurenz Albe wrote: > One reason could be index corruption. If one query uses an index and the > other doesn't, > that could lead to different results. > > The other option is of course a trivial

if not exists (SELECT 1... UNION SELECT 1...)

Hello, when I search for a non existent word in the two tables hosted in PostgreSQL 14.1 then I get zero records as expected: words_en=> SELECT 1 FROM words_nouns WHERE word = 'ABCDE' UNION SELECT 1 FROM words_verbs WHERE word = 'ABCDE'; ?column? -- (0 rows) But when I try to use the

Re: Stored function RETURNS table, but in some cases columns are missing - should I set them to NULL?

Tom, you are so eagle eyed - On Mon, Mar 8, 2021 at 8:53 PM Tom Lane wrote: > Alexander Farber writes: > > However there are cases, when I only have the out_gid value, I do not > want > > to return any other values. > > My question is: do I have to set the other OUT p

Stored function RETURNS table, but in some cases columns are missing - should I set them to NULL?

Good evening, in PostgreSQL 13.2 I have a custom stored function: CREATE OR REPLACE FUNCTION words_join_new_game( in_uid integer, in_bid integer ) RETURNS table ( -- the player to be notified (sometimes there is no such user)

Re: Localizing stored functions by replacing placeholders in their body

Thanks for your input Actually, yes, that is what I have right now a translate() like stored function, with format %s sometimes. But that is "at runtime" and I would like to have a "at compile time"/"deploy once and forget" solution, that is why I have asked about approaches for modifying the

Re: Localizing stored functions by replacing placeholders in their body

I think I will try this approach: \set localized_declaration `sed 's/this/that/' my_func.sql` :localized_declaration Thank you for your input

Re: Localizing stored functions by replacing placeholders in their body

Ah, I understand, that was the wrong EXECUTE, thank you. Another idea: can't I use \set command for my purpose of localizing stored functions? \set my_func_declaration `sed 's/this/that/' my_func.sql` But how to execute the declaration? I can only echo it with select (:'my_func_declaration');

Re: Localizing stored functions by replacing placeholders in their body

Thank you for the \! hint, Pavel, didn't know about that! Is it possible to have a pure SQL solution? (To avoid having to install "sed" on my Win 10 PC) Maybe by using EXECUTE? EXECUTE REGEXP_REPLACE( $localize$ CREATE OR REPLACE FUNCTION my_func() RETURNS text AS $func$ SELECT

Re: Localizing stored functions by replacing placeholders in their body

Or is it possible to call external commands from an sql script, like \i "sed 's/this/that/' some.sql"

Re: Localizing stored functions by replacing placeholders in their body

Yes, good point about the '\$', thank you Tom. The reason I am trying not to use sed, is because I deploy my database by executing a single command: psql words_en < words_en.sql And the file words_en.sql has the contents: \i words_hash.sql \i words_all_letters.sql \i words_get_hint.sql \i

Localizing stored functions by replacing placeholders in their body

Hello, I have an app using PostgreSQL 13.2, in 6 different human languages (each using different database, but same source code). Currently to localize strings return/set by the stored functions I either get localized strings from a table or maintain stored function source code in 6 different

Re: Deleting takes days, should I add some index?

Thank you for the explanation, David On Thu, Feb 25, 2021 at 9:49 PM David Rowley wrote: > > Since your foreign keys perform a cascade delete on the tables > referencing the tables you're deleting from, any records in those > referencing tables will be deleted too. You must also look at those

Re: Deleting takes days, should I add some index?

Thank you, Pavel! I didn't even think about trying to "explain analyze" deletion of just 1 record - On Thu, Feb 25, 2021 at 10:04 PM Pavel Stehule wrote: > čt 25. 2. 2021 v 19:39 odesílatel Alexander Farber < >> alexander.far...@gmail.com> napsal: >> >>&

Re: Deleting takes days, should I add some index?

Pavel, thank you for asking! I have put the anonymized dump of my database at: http://wordsbyfarber.com/words_dev.sql.gz (beware, it is a 1.3 GB download) The question is why does the command take days (when I tried last time): delete from words_games where created < now() - interval '12

Re: Deleting takes days, should I add some index?

Hi Pavel, trying to follow your advice "You should check so all foreign keys have an index" I look at the table where I want to delete older records: # \d words_games Table "public.words_games" Column | Type | Collation | Nullable |

Re: Deleting takes days, should I add some index?

Hello, revisiting an older mail on the too long deletion times (in PostgreSQL 13.2)... I have followed the advices here, thank you - On Fri, Nov 27, 2020 at 4:15 PM Guillaume Lelarge wrote: > Le ven. 27 nov. 2020 à 16:05, Alvaro Herrera a > écrit : > >> On 2020-Nov-27, Alexand

Re: JSONB_AGG: aggregate function calls cannot be nested

Ahh, thank you all - select row_to_json (x) FROM( SELECT jsonb_agg(day) AS day, jsonb_agg(completed) AS completed, jsonb_agg(expired) AS expired from ( SELECT TO_CHAR(finished, '-MM-DD') AS day, count(*) filter (where reason in ('regular', 'resigned'))

Re: JSONB_AGG: aggregate function calls cannot be nested

Thank you Thomas, this results in select day AS day, jsonb_agg(completed) AS completed, jsonb_agg(expired) AS expired from ( SELECT TO_CHAR(finished, '-MM-DD') AS day, count(*) filter (where reason in ('regular', 'resigned')) AS completed,

Re: JSONB_AGG: aggregate function calls cannot be nested

Ah, thank you... JSON support in PostgreSQL is cool and seems to be extended with each release. But standard tasks of returning a JSON map of lists or JSON list of list seem to be difficult to use. Greetings Alex

Re: JSONB_AGG: aggregate function calls cannot be nested

Then I have to split the query in 3 similar ones (with same condition)? I try: SELECT JSONB_AGG(TO_CHAR(finished, '-MM-DD')) AS day FROM words_games WHERE finished > CURRENT_TIMESTAMP - INTERVAL '2 week' GROUP BY day ORDER BY day; ERROR:

JSONB_AGG: aggregate function calls cannot be nested

Good evening, I have the following query in 13.2: # SELECT TO_CHAR(finished, '-MM-DD') AS day, SUM(CASE WHEN reason='regular' or reason='resigned' THEN 1 ELSE 0 END)::int AS completed, SUM(CASE WHEN reason='expired' THEN 1 ELSE 0 END)::int AS

Re: Please help to improve SELECT JOIN performance, EXPLAIN ANALYZE attached

: logical replication launcher On Sat, Feb 20, 2021 at 4:15 PM Alexander Farber wrote: > Thank you all, I will try at first > > shared_buffers = 16GBand > > index on words_scores(uid, length(word) desc) > > >

Re: Please help to improve SELECT JOIN performance, EXPLAIN ANALYZE attached

Thank you all, I will try at first shared_buffers = 16GBand index on words_scores(uid, length(word) desc)

Re: Please help to improve SELECT JOIN performance, EXPLAIN ANALYZE attached

Yes, Michael, that I have noticed too, but should have written more in my original mail. The query when I try it does run in 15ms, but evening logs show the query (I think only small percentage of it) running 1-3s. At the same time my CentOS 8 server with 64 GB RAM is never loaded, the load

Please help to improve SELECT JOIN performance, EXPLAIN ANALYZE attached

Good evening, I have a word game which uses PostgreSQL 13.2 and 80% of the code is written as stored functions in PL/PgSQL or SQL. Recently I have purchased some traffic and the number of daily games increased from 100 to 700. In the PostgreSQL log I have noticed that the duration for 2

Re: How to return a jsonb list of lists (with integers)

I have tried switching to SELECT INTO, but IF FOUND is still always true, which gives me [ null, null, null ] for some users: SELECT JSONB_BUILD_ARRAY( SUM(CASE WHEN (player1 = in_viewer AND state1 = 'won') OR (player2 = in_viewer AND state2 = 'won') THEN

Re: How to return a jsonb list of lists (with integers)

Hi Pavel, why would SELECT INTO be better here? Thanks Alex

Re: How to return a jsonb list of lists (with integers)

I have ended up with the following (to avoid returning [null, null, null] for players who never played with each other): _versus := JSONB_BUILD_ARRAY( SUM(CASE WHEN (player1 = in_uid AND state1 = 'won') OR (player2 = in_uid AND state2 = 'won') THEN 1 ELSE

Re: How to return a jsonb list of lists (with integers)

I have tried the following, but IF FOUND is always false for some reason: _versus := JSONB_BUILD_ARRAY( SUM(CASE WHEN (player1 = in_uid AND state1 = 'won') OR (player2 = in_uid AND state2 = 'won') THEN 1 ELSE 0 END)::integer,

Re: How to return a jsonb list of lists (with integers)

Hello, thank you for the helpful replies. I have decided to go with PL/PgSQL for now and also switched from JSONB list of lists to map of lists. And the custom stored function below works mostly well, except for a special case - CREATE OR REPLACE FUNCTION words_stat_charts(

Re: How to return a jsonb list of lists (with integers)

Thank you, David, with json_build_array() it works for a single query - SELECT JSONB_BUILD_ARRAY( SUM(CASE WHEN (player1 = in_uid AND state1 = 'won') OR (player2 = in_uid AND state2 = 'won') THEN 1 ELSE 0 END)::integer, SUM(CASE WHEN (player1

Re: How to return a jsonb list of lists (with integers)

On Tue, Feb 16, 2021 at 7:52 PM Michael Lewis wrote: > Aggregate functions work on a single column to summarize many rows into > fewer rows. You seem to be wanting to combine multiple columns which would > be done by concatenation or array[column1,column2] or something like that. > Ah right,

How to return a jsonb list of lists (with integers)

Good evening, In 13.2 I have 3 SQL queries, which work well and return integer values. The values I feed to Google Charts (and currently I switch to Chart.js). Currently I use the queries by calling 3 different custom stored functions by my Java servlet. I would like to convert the functions

Re: Select a column and then apply JSONB_ARRAY_ELEMENTS to it

On Sat, Jan 9, 2021 at 3:49 PM David G. Johnston wrote: > Subqueries must be enclosed in parentheses. The parentheses that are part > of the function call do not count. > > Ah! Thank you David, this has worked now - CREATE OR REPLACE FUNCTION words_solve_puzzle( in_mid

Select a column and then apply JSONB_ARRAY_ELEMENTS to it

Hello, for a word puzzle using PostgreSQL 13.1: https://wortefarbers.de/ws/puzzle2?mid=138=c6f469786df7e8d44461381b62b2ce7d I am trying to improve a stored function - CREATE OR REPLACE FUNCTION words_solve_puzzle( in_mid bigint, in_uid int,

Re: SELECT but only if not present in another table

Thank you, Steve - On Sun, Dec 6, 2020 at 6:50 PM Steve Baldwin wrote: > Can't you just use table aliases? So, the outer word_moves would become > 'word_moves as wm', word_puzzles would become 'word_puzzles as wp', and the > where clause 'WHERE wp.mid = wm.mid' ? > table aliases have worked

SELECT but only if not present in another table

Good evening, in PostgreSQL 13.1 I save player moves in the table: # \d words_moves Table "public.words_moves" Column | Type | Collation | Nullable | Default

Re: Using a boolean column with IF / THEN

On Sat, Dec 5, 2020 at 9:00 PM David G. Johnston wrote: > Maybe not “simpler” but for all those checks you could write a single > query that pulls out all the data at once into a record variable and test > against the columns pf that instead of executing multiple queries. > Thank you!

Using a boolean column with IF / THEN

Good evening, hopefully my question is not too stupid, but - in a 13.1 database I have a words_users table with a boolean column: -- the user is not allowed to chat or change the motto muted boolean NOT NULL DEFAULT false, Currently I check the value as follows, but I

Deleting takes days, should I add some index?

Hello, I am using PostgreSQL 10.15 on CentOS 7 with 64 GB RAM, Intel i7 6700 and I have the following 2 tables there: words_ru=> \d words_games Table "public.words_games" Column | Type | Collation | Nullable | Default

Re: How to call JSONB_INSERT with integer as the new to-be-inserted value?

Thank you!

How to call JSONB_INSERT with integer as the new to-be-inserted value?

Good evening, I am trying to take a JSONB object (comes from an HTTP cookie set by my app) and add a property "uid" to it, which should hold an integer: CREATE OR REPLACE FUNCTION words_get_user( in_users jsonb, OUT out_user jsonb ) RETURNS jsonb AS $func$

CREATE INDEX ON words_moves(length(letters), score) WHERE action = 'play'

Good evening, in PostgreSQL 11 I have a table holding player moves (could be: 'play', 'swap', 'skip', ...) in a word game: # \d words_moves; Table "public.words_moves" Column | Type | Collation | Nullable | Default

Re: Syntax error for UPDATE ... RETURNING INTO STRICT

Thanks for your replies! Tom has hinted that STRICT is pl/pgSQL syntax and not SQL Regards Alex >

Re: Syntax error for UPDATE ... RETURNING INTO STRICT

Thank you Patrick - On Tue, Dec 3, 2019 at 11:49 AM Patrick FICHE wrote: > > It seems that STRICT is the issue. > > But why does your function return a table of boolean in this case ? > > As it only updates one record, it would probably be easier to return a > boolean only. > > CREATE OR

Syntax error for UPDATE ... RETURNING INTO STRICT

Good morning, why does not PostgreSQL 10.11 please like the - CREATE OR REPLACE FUNCTION words_toggle_puzzle( in_mid bigint ) RETURNS table ( out_puzzle boolean ) AS $func$ UPDATE words_moves SET puzzle = NOT puzzle

Re: Counting booleans in GROUP BY sections

My context is that I have a table of player moves with PK mid (aka "move id"). And I am able to find "interesting" moves by the high score or all 7 letter tiles used. But I do some human reviewing and set a "puzzle" boolean for truly interesting moves. For the reviewing tool I would like to

Re: Counting booleans in GROUP BY sections

Thank you Adrian, but - On Fri, Nov 29, 2019 at 6:45 PM Adrian Klaver wrote: > On 11/29/19 8:38 AM, Alexander Farber wrote: > > > > CREATE OR REPLACE FUNCTION words_list_puzzles( > > in_start interval, > > in_end interval >

Counting booleans in GROUP BY sections

Good evening, I am trying to count the booleans per each GROUP BY section by the following stored function: CREATE OR REPLACE FUNCTION words_list_puzzles( in_start interval, in_end interval ) RETURNS TABLE ( out_label text,

Re: Trying to fetch records only if preceded by at least another one

Thank you Michael - On Fri, Oct 25, 2019 at 7:28 PM Michael Lewis wrote: > But it prints too many records: all moves cross-multiplied with each other. >> >> As if I have forgotten to add 1 more condition to the JOIN LATERAL >> > > LIMIT 1 inside your lateral should resolve that. Personally, I'd

Trying to fetch records only if preceded by at least another one

Good evening, for a word game hosted on PostgreSQL 10 I try to find interesting player moves (high score or played all 7 tiles) and generate a "puzzle" images out of them (example: https://imgur.com/a/StnXqoR ) The moves are stored in: words_ru=> \d words_moves

Re: Calling jsonb_array_elements 4 times in the same query

Thank you - On Mon, Oct 21, 2019 at 11:20 PM Adrian Klaver wrote: > As Thomas pointed there is a difference between -> and ->>: > > test_(postgres)# select pg_typeof('[{"one": 1, "two": 2}]'::jsonb -> 0 > -> 'one'), '[{"one": 1, "two": 2}]'::jsonb -> 0 -> 'one'; > pg_typeof | ?column? >

Re: Calling jsonb_array_elements 4 times in the same query

Apologies, I should have shown the JSON structure in my very first email - On Mon, Oct 21, 2019 at 4:45 PM Thomas Kellerer wrote: > Use ->> to return the value as text (not as JSONB) and you need to use the > column alias, not the table alias: > > (t.tile ->> 'col')::int > > It is a

Re: Calling jsonb_array_elements 4 times in the same query

Thank you Thomas - On Mon, Oct 21, 2019 at 4:24 PM Thomas Kellerer wrote: > Alexander Farber schrieb am 21.10.2019 um 15:39: > > I am trying to construct a query, which would draw a game board when > given a move id (aka mid): > > > >

Calling jsonb_array_elements 4 times in the same query

Hello, good afternoon! With PostgreSQL 10 I host a word game, which stores player moves as a JSON array of objects with properties: col, row, value, letter - CREATE TABLE words_moves ( mid BIGSERIAL PRIMARY KEY, action text NOT NULL, gid integer NOT NULL

Re: Azure Database for PostgreSQL - how to add pgbouncer or similar?

Thank you Adrian - On Sat, Jun 29, 2019 at 3:59 PM Adrian Klaver wrote: > On 6/29/19 12:34 AM, Alexander Farber wrote: > > Not from within. Supported extensions: > > https://docs.microsoft.com/en-us/azure/postgresql/concepts-extensions > > Using search term 'azure postgre

Azure Database for PostgreSQL - how to add pgbouncer or similar?

Good morning, in Microsoft Azure how to add pgbouncer or a similar software enabling multiple simultaneous connections? After reading https://docs.microsoft.com/en-us/azure/postgresql/overview I could not find such options... Thank you Alex

Re: How to return ARRAY from SQL function?

Thank you - On Tue, Jun 18, 2019 at 3:10 PM Tom Lane wrote: > Alexander Farber writes: > > And then I shuffle the letters by - > > > CREATE OR REPLACE FUNCTION words_shuffle(in_array text[]) > > RETURNS text[] AS > > $func$ > > SELECT array_

Re: How to return ARRAY from SQL function?

Thank you, Laurenz and Tom - On Fri, Jun 14, 2019 at 3:25 PM Tom Lane wrote: > > Laurenz Albe writes: > > > You'll have to specify an array of which type you want, probably > > ... RETURNS text[] > > Right. Also, I don't recall the exact rules in this area, but I think > that SQL functions

How to return ARRAY from SQL function?

Hello, in PostgreSQL 10.8 the following works - words_ru=> SELECT ARRAY[ words_ru-> '*', '*', 'А', 'А', 'А', 'А', 'А', 'А', 'А', 'А', words_ru-> 'Б', 'Б', 'В', 'В', 'В', 'В', 'Г', 'Г', 'Д', 'Д', words_ru-> 'Д', 'Д', 'Е', 'Е', 'Е', 'Е', 'Е',

Re: aggregate functions are not allowed in UPDATE

Thank you, the following seems to have worked - On Tue, Jan 15, 2019 at 8:49 PM David G. Johnston < david.g.johns...@gmail.com> wrote: > > UPDATE users > SET avg_time = diffs.average_time_for_the_grouped_by_user > FROM diffs > WHERE users.uid = diffs.uid --< the missing "where" I commented about

Re: aggregate functions are not allowed in UPDATE

Last question please - how to run the query for all users? I know I could use the FOR-loop from PL/PgSQL, but is there also a pure SQL way? How to refer to the outside "uid" from inside the CTE in the query below? WITH diffs AS ( SELECT gid, uid, played - LAG(played)

Re: aggregate functions are not allowed in UPDATE

And I should better change the avg_time column from TEXT to TIMESTAMPTZ (and use TO_CHAR on it later down the road) so that I can compare my players Regards Alex >

Re: aggregate functions are not allowed in UPDATE

Ahh, the subqueries - On Tue, Jan 15, 2019 at 5:59 PM David G. Johnston < david.g.johns...@gmail.com> wrote: > On Tue, Jan 15, 2019 at 9:52 AM Alexander Farber > wrote: > >> So calculate the average somewhere else, put the result in a column, > >> and then re

Re: aggregate functions are not allowed in UPDATE

Unfortunately, I don't understand your advice, David - On Tue, Jan 15, 2019 at 5:46 PM David G. Johnston < david.g.johns...@gmail.com> wrote: > On Tue, Jan 15, 2019 at 9:42 AM Alexander Farber > wrote: > > When I am trying > > > > WITH diffs AS ( > >

aggregate functions are not allowed in UPDATE

Good evening, I have prepared a simple test case for my question: https://www.db-fiddle.com/f/w1AYGpoZiWW9bLCYjHDk7H/0 There I create 3 tables: CREATE TABLE users ( uid SERIAL PRIMARY KEY, avg_time TEXT ); CREATE TABLE games ( gid SERIAL PRIMARY KEY, player1 INTEGER NOT NULL REFERENCES

Re: How to always run UPDATE FROM despite missing records in the source table?

Thank you Adrian - On Fri, Jan 11, 2019 at 4:55 PM Adrian Klaver wrote: > On 1/11/19 4:50 AM, Alexander Farber wrote: > > https://www.db-fiddle.com/f/22jfWnsvqD8hVeFPXsyLbV/0 > > Why not put a test for the block in the function and then use different > UPDATE's depending o

How to always run UPDATE FROM despite missing records in the source table?

Good afternoon I have prepared a simplified test case for my question: https://www.db-fiddle.com/f/22jfWnsvqD8hVeFPXsyLbV/0 In PostgreSQL 10.6 there are 2 tables: CREATE TABLE users ( uid SERIAL PRIMARY KEY, created timestamptz NOT NULL, visited timestamptz

Re: Not sure which part of the query needs optimization

Thank you for the comments, Andrew - On Mon, Jan 7, 2019 at 8:40 PM Andrew Gierth wrote: > The obvious thing to do is to keep a computed average score for each > user - either in a separate table which you update based on changes to > words_moves, which you could do with a trigger, or using a

Re: Adding LEFT JOIN to a query has increased execution time 10 times

Hi Andrew - On Mon, Jan 7, 2019 at 12:00 AM Andrew Gierth wrote: > >>>>> "Alexander" == Alexander Farber writes: > Alexander> With the further help of the IRC folks the query has been > Alexander> optimized (7-10 seconds -> 0.3 second) >

Not sure which part of the query needs optimization

Good afternoon, for each visitor of my website I generate a JSON list of 30 top players ( https://slova.de/words/top.php ), who played in the past week, with their average scores and average time between moves. With 5 seconds this query is taking quite a bit of time:

Re: Adding LEFT JOIN to a query has increased execution time 10 times

Good evening, thank you for the useful hints! With the further help of the IRC folks the query has been optimized (7-10 seconds -> 0.3 second) by adding the following indices: CREATE INDEX ON words_games (player1, COALESCE(finished, 'INFINITY')); CREATE INDEX ON words_games (player2,

Re: Adding LEFT JOIN to a query has increased execution time 10 times

Oh ok, so it is not as simple as eliminating all "Seq Scan" occurrences... Thank you for replying Andrew - On Sat, Jan 5, 2019 at 9:18 PM Andrew Gierth wrote: That seems slow in itself, even before adding the extra join - the > explain suggests that you're both short on indexes and you're

Adding LEFT JOIN to a query has increased execution time 10 times

Good evening, On a CentOS 7.6 server (Intel Core i7-6700, 64 GB DDR4 RAM, RAID1 SSD) I run a backend written in PL/pgSQL and Java for a mobile and desktop word game with the following Linux packages: postgresql10-server-10.6-1PGDG.rhel7.x86_64 pgbouncer-1.9.0-1.rhel7.x86_64

SELECT UNION into a text[]

Good evening, I am trying to SELECT ARRAY_AGG into an array from 2 tables. But unfortunately here is what I get in PostgreSQL 10.5: SELECT ARRAY_AGG(hashed) FROM words_nouns WHERE added > TO_TIMESTAMP(1539100913) UNION SELECT ARRAY_AGG(hashed) FROM

Re: UPDATE from CTE syntax error

On Sat, May 26, 2018 at 6:37 PM, Adrian Klaver wrote: > > > https://www.postgresql.org/docs/10/static/sql-update.html > "column_name > > The name of a column in the table named by table_name. The column name > can be qualified with a subfield name or array

Re: UPDATE from CTE syntax error

Oops, I am sorry for the formatting - Mac + Terminal + Gmail :-/

  1   2   >