From 18a2e8e4d8da54a0dc498578c2c20c358d6db49d Mon Sep 17 00:00:00 2001
From: stedif <stefano.difrischia@inaf.it>
Date: Thu, 3 Mar 2022 18:04:02 +0100
Subject: [PATCH] L2SS-647: add timescaledb extensions and bug fix
---
docker-compose/timescaledb/Dockerfile | 25 +-
.../resources/04_hdb_ext_users.sql | 32 +
...gregates.sql => 05_hdb_ext_aggregates.sql} | 1 +
.../06_hdb_ext_arrays_aggregates_helper.sql | 2336 +++++++++++++++++
.../07_hdb_ext_arrays_aggregates.sql | 328 +++
.../resources/08_hdb_ext_compress_policy.sql | 118 +
.../resources/09_hdb_ext_import.sql | 19 +
.../resources/10_hdb_ext_reorder_policy.sql | 53 +
.../{05_lofar_func.sh => 11_lofar_func.sh} | 0
...{06_lofar_views.sql => 12_lofar_views.sql} | 0
.../{07_cleanup.sql => 13_cleanup.sql} | 0
11 files changed, 2907 insertions(+), 5 deletions(-)
create mode 100644 docker-compose/timescaledb/resources/04_hdb_ext_users.sql
rename docker-compose/timescaledb/resources/{04_hdb_ext_aggregates.sql => 05_hdb_ext_aggregates.sql} (99%)
create mode 100644 docker-compose/timescaledb/resources/06_hdb_ext_arrays_aggregates_helper.sql
create mode 100644 docker-compose/timescaledb/resources/07_hdb_ext_arrays_aggregates.sql
create mode 100644 docker-compose/timescaledb/resources/08_hdb_ext_compress_policy.sql
create mode 100644 docker-compose/timescaledb/resources/09_hdb_ext_import.sql
create mode 100644 docker-compose/timescaledb/resources/10_hdb_ext_reorder_policy.sql
rename docker-compose/timescaledb/resources/{05_lofar_func.sh => 11_lofar_func.sh} (100%)
rename docker-compose/timescaledb/resources/{06_lofar_views.sql => 12_lofar_views.sql} (100%)
rename docker-compose/timescaledb/resources/{07_cleanup.sql => 13_cleanup.sql} (100%)
diff --git a/docker-compose/timescaledb/Dockerfile b/docker-compose/timescaledb/Dockerfile
index f97faca9a..8d250ae85 100644
--- a/docker-compose/timescaledb/Dockerfile
+++ b/docker-compose/timescaledb/Dockerfile
@@ -1,10 +1,25 @@
FROM timescale/timescaledb:latest-pg12
-
+# Set admin role to perform DB creation
COPY resources/01_admin.sh docker-entrypoint-initdb.d/002_admin.sh
+# Create DB schema (tables, indexes, etc.)
COPY resources/02_hdb_schema.sql docker-entrypoint-initdb.d/003_hdb_schema.sql
+# Create DB roles
COPY resources/03_hdb_roles.sql docker-entrypoint-initdb.d/004_hdb_roles.sql
-COPY resources/04_hdb_ext_aggregates.sql docker-entrypoint-initdb.d/005_hdb_ext_aggregates.sql
-COPY resources/05_lofar_func.sh docker-entrypoint-initdb.d/006_lofar_func.sh
-COPY resources/06_lofar_views.sql docker-entrypoint-initdb.d/007_lofar_views.sql
-COPY resources/07_cleanup.sql docker-entrypoint-initdb.d/008_cleanup.sql
+# Create further roles
+COPY resources/04_hdb_ext_users.sql docker-entrypoint-initdb.d/005_hdb_ext_users.sql
+# Create timescaledb aggregates
+COPY resources/05_hdb_ext_aggregates.sql docker-entrypoint-initdb.d/006_hdb_ext_aggregates.sql
+COPY resources/06_hdb_ext_arrays_aggregates_helper.sql docker-entrypoint-initdb.d/007_hdb_ext_arrays_aggregates_helper.sql
+COPY resources/07_hdb_ext_arrays_aggregates.sql docker-entrypoint-initdb.d/008_hdb_ext_arrays_aggregates.sql
+# Add compress policy
+COPY resources/08_hdb_ext_compress_policy.sql docker-entrypoint-initdb.d/009_hdb_ext_compress_policy.sql
+# Add further functions
+COPY resources/09_hdb_ext_import.sql docker-entrypoint-initdb.d/010_hdb_ext_import.sql
+# Add reorder policy
+COPY resources/10_hdb_ext_reorder_policy.sql docker-entrypoint-initdb.d/011_hdb_ext_reorder_policy.sql
+# Add LOFAR functions and views
+COPY resources/11_lofar_func.sh docker-entrypoint-initdb.d/012_lofar_func.sh
+COPY resources/12_lofar_views.sql docker-entrypoint-initdb.d/013_lofar_views.sql
+# Cleanup admin role
+COPY resources/13_cleanup.sql docker-entrypoint-initdb.d/014_cleanup.sql
diff --git a/docker-compose/timescaledb/resources/04_hdb_ext_users.sql b/docker-compose/timescaledb/resources/04_hdb_ext_users.sql
new file mode 100644
index 000000000..1ce744e86
--- /dev/null
+++ b/docker-compose/timescaledb/resources/04_hdb_ext_users.sql
@@ -0,0 +1,32 @@
+-- -----------------------------------------------------------------------------
+-- This file is part of the hdbpp-timescale-project
+--
+-- Copyright (C) : 2014-2019
+-- European Synchrotron Radiation Facility
+-- BP 220, Grenoble 38043, FRANCE
+--
+-- libhdb++timescale is free software: you can redistribute it and/or modify
+-- it under the terms of the Lesser GNU General Public License as published by
+-- the Free Software Foundation, either version 3 of the License, or
+-- (at your option) any later version.
+--
+-- libhdb++timescale is distributed in the hope that it will be useful,
+-- but WITHOUT ANY WARRANTY; without even the implied warranty of
+-- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Lesser
+-- GNU General Public License for more details.
+--
+-- You should have received a copy of the Lesser GNU General Public License
+-- along with libhdb++timescale. If not, see <http://www.gnu.org/licenses/>.
+-- -----------------------------------------------------------------------------
+
+\c hdb
+
+-- Some useful users for a basic system
+CREATE ROLE hdb_cfg_man WITH LOGIN PASSWORD 'hdbpp';
+GRANT readwrite TO hdb_cfg_man;
+
+CREATE ROLE hdb_event_sub WITH LOGIN PASSWORD 'hdbpp';
+GRANT readwrite TO hdb_event_sub;
+
+CREATE ROLE hdb_data_reporter WITH LOGIN PASSWORD 'hdbpp';
+GRANT readonly TO hdb_data_reporter;
diff --git a/docker-compose/timescaledb/resources/04_hdb_ext_aggregates.sql b/docker-compose/timescaledb/resources/05_hdb_ext_aggregates.sql
similarity index 99%
rename from docker-compose/timescaledb/resources/04_hdb_ext_aggregates.sql
rename to docker-compose/timescaledb/resources/05_hdb_ext_aggregates.sql
index 7c3484299..a253ccd04 100644
--- a/docker-compose/timescaledb/resources/04_hdb_ext_aggregates.sql
+++ b/docker-compose/timescaledb/resources/05_hdb_ext_aggregates.sql
@@ -1290,3 +1290,4 @@ GRANT SELECT ON cagg_scalar_devushort_1day TO readonly;
-- DROP VIEW cagg_scalar_devushort_1hour CASCADE;
-- DROP VIEW cagg_scalar_devushort_8hour CASCADE;
-- DROP VIEW cagg_scalar_devushort_1day CASCADE;
+
diff --git a/docker-compose/timescaledb/resources/06_hdb_ext_arrays_aggregates_helper.sql b/docker-compose/timescaledb/resources/06_hdb_ext_arrays_aggregates_helper.sql
new file mode 100644
index 000000000..7c67f6536
--- /dev/null
+++ b/docker-compose/timescaledb/resources/06_hdb_ext_arrays_aggregates_helper.sql
@@ -0,0 +1,2336 @@
+-- -----------------------------------------------------------------------------
+-- This file is part of the hdbpp-timescale-project
+--
+-- Copyright (C) : 2014-2019
+-- European Synchrotron Radiation Facility
+-- BP 220, Grenoble 38043, FRANCE
+--
+-- libhdb++timescale is free software: you can redistribute it and/or modify
+-- it under the terms of the Lesser GNU General Public License as published by
+-- the Free Software Foundation, either version 3 of the License, or
+-- (at your option) any later version.
+--
+-- libhdb++timescale is distributed in the hope that it will be useful,
+-- but WITHOUT ANY WARRANTY; without even the implied warranty of
+-- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Lesser
+-- GNU General Public License for more details.
+--
+-- You should have received a copy of the Lesser GNU General Public License
+-- along with libhdb++timescale. If not, see <http://www.gnu.org/licenses/>.
+-- -----------------------------------------------------------------------------
+
+-- Aggregates function helper for the continuous aggregates views for the array attributes.
+
+\c hdb
+
+/*
+NOT USED
+Keep as a different approach, should be benchmarked for execution speed
+
+-- Special type to be used as input by compute_element_agg
+-- It contains past aggregates result and the new values
+create type double_agg_input as (
+ value_r double precision,
+ value_w double precision,
+ count_r integer,
+ count_nan_r integer,
+ avg_r decimal,
+ min_r double precision,
+ max_r double precision,
+ stddev_r decimal,
+ count_w integer,
+ count_nan_w integer,
+ avg_w decimal,
+ min_w double precision,
+ max_w double precision,
+ stddev_w decimal
+);
+
+-- Function to compute the aggregates from the new values and old aggregates
+-- result.
+-- It computes the result for an array of input and return a table so that it
+-- can be used in a FROM clause
+CREATE OR REPLACE FUNCTION compute_element_agg(inp_arr double_agg_input[]
+ ) RETURNS SETOF RECORD as $$
+
+DECLARE
+ ret RECORD;
+ inp double_agg_input;
+ value_r double precision;
+ value_w double precision;
+ count_r integer;
+ count_nan_r integer;
+ avg_r decimal;
+ min_r double precision;
+ max_r double precision;
+ stddev_r decimal;
+ count_w integer;
+ count_nan_w integer;
+ avg_w decimal;
+ min_w double precision;
+ max_w double precision;
+ stddev_w decimal;
+ n_count_r integer;
+ n_count_nan_r integer;
+ n_avg_r decimal;
+ n_min_r double precision;
+ n_max_r double precision;
+ n_stddev_r decimal;
+ n_count_w integer;
+ n_count_nan_w integer;
+ n_avg_w decimal;
+ n_min_w double precision;
+ n_max_w double precision;
+ n_stddev_w decimal;
+
+BEGIN
+ FOREACH inp IN ARRAY inp_arr
+ LOOP
+
+ value_r := inp.value_r;
+ value_w := inp.value_w;
+ count_r := inp.count_r;
+ count_nan_r := inp.count_nan_r;
+ avg_r := inp.avg_r;
+ min_r := inp.min_r;
+ max_r := inp.max_r;
+ stddev_r := inp.stddev_r;
+ count_w := inp.count_w;
+ count_nan_w := inp.count_nan_w;
+ avg_w := inp.avg_w;
+ min_w := inp.min_w;
+ stddev_w := inp.stddev_w;
+
+ IF value_r IS NULL OR value_r='NaN'::float8 OR value_r='Infinity' OR value_r='-Infinity'
+ THEN
+
+ IF count_r IS NULL
+ THEN
+ n_count_r = 0;
+ ELSE
+ n_count_r = count_r;
+ END IF;
+
+ IF value_r IS NULL
+ THEN
+
+ IF count_nan_r IS NULL
+ THEN
+ n_count_nan_r = 0;
+ ELSE
+ n_count_nan_r = count_nan_r;
+ END IF;
+
+ ELSE
+
+ IF count_nan_r IS NULL
+ THEN
+ n_count_nan_r = 1;
+ ELSE
+ n_count_nan_r = count_nan_r + 1;
+ END IF;
+ END IF;
+
+ n_avg_r = avg_r;
+ n_min_r = min_r;
+ n_max_r = max_r;
+ n_stddev_r = stddev_r;
+
+ ELSE
+
+ IF count_nan_r IS NULL
+ THEN
+ n_count_nan_r = 0;
+ ELSE
+ n_count_nan_r = count_nan_r;
+ END IF;
+
+ IF count_r IS NULL
+ THEN
+ n_count_r = 1;
+ ELSE
+ n_count_r = count_r + 1;
+ END IF;
+
+ IF avg_r IS NULL
+ THEN
+ n_avg_r = value_r;
+ ELSE
+ n_avg_r = avg_r + (value_r-avg_r)/(count_r+1.)::decimal;
+ END IF;
+
+ n_min_r = LEAST(value_r, min_r);
+ n_max_r = GREATEST(value_r, max_r);
+
+ IF stddev_r IS NULL
+ THEN
+ n_stddev_r = 0;
+ ELSE
+ n_stddev_r = stddev_r + ((count_r + 0.)/(count_r+1.))*power(value_r - avg_r, 2);
+ END IF;
+ END IF;
+
+ IF value_w IS NULL OR value_w='NaN'::float8 OR value_w='Infinity' OR value_w='-Infinity'
+ THEN
+
+ IF count_w IS NULL
+ THEN
+ n_count_w = 0;
+ ELSE
+ n_count_w = count_w;
+ END IF;
+
+ IF value_w IS NULL
+ THEN
+
+ IF count_nan_w IS NULL
+ THEN
+ n_count_nan_w = 0;
+ ELSE
+ n_count_nan_w = count_nan_w;
+ END IF;
+
+ ELSE
+
+ IF count_nan_w IS NULL
+ THEN
+ n_count_nan_w = 1;
+ ELSE
+ n_count_nan_w = count_nan_w + 1;
+ END IF;
+ END IF;
+
+ n_avg_w = avg_w;
+ n_min_w = min_w;
+ n_max_w = max_w;
+ n_stddev_w = stddev_w;
+
+ ELSE
+
+ IF count_nan_w IS NULL
+ THEN
+ n_count_nan_w = 0;
+ ELSE
+ n_count_nan_w = count_nan_w;
+ END IF;
+
+ IF count_w IS NULL
+ THEN
+ n_count_w = 1;
+ ELSE
+ n_count_w = count_w + 1;
+ END IF;
+
+ IF avg_w IS NULL
+ THEN
+ n_avg_w = value_w;
+ ELSE
+ n_avg_w = avg_w + (value_w-avg_w)/(count_w+1);
+ END IF;
+
+ n_min_w = LEAST(value_w, min_w);
+ n_max_w = GREATEST(value_w, max_w);
+
+ IF stddev_w IS NULL
+ THEN
+ n_stddev_w = 0;
+ ELSE
+ n_stddev_w = stddev_w + ((count_w + 0.)/(count_w+1.)*power(value_w - avg_w, 2));
+ END IF;
+ END IF;
+
+ ret := (n_count_r, n_count_nan_r, n_avg_r, n_min_r, n_max_r, n_stddev_r
+ , n_count_w, n_count_nan_w, n_avg_w, n_min_w, n_max_w, n_stddev_w);
+
+ return next ret;
+ END LOOP;
+END;
+$$
+LANGUAGE 'plpgsql';
+*/
+
+
+-- Special types to store the aggregations data during computation
+create type double_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ count_nan_r integer[],
+ avg_r decimal[],
+ min_r double precision[],
+ max_r double precision[],
+ stddev_r decimal[],
+ count_w integer[],
+ count_nan_w integer[],
+ avg_w decimal[],
+ min_w double precision[],
+ max_w double precision[],
+ stddev_w decimal[]
+);
+
+create type float_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ count_nan_r integer[],
+ avg_r decimal[],
+ min_r real[],
+ max_r real[],
+ stddev_r decimal[],
+ count_w integer[],
+ count_nan_w integer[],
+ avg_w decimal[],
+ min_w real[],
+ max_w real[],
+ stddev_w decimal[]
+);
+
+create type long_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ avg_r decimal[],
+ min_r integer[],
+ max_r integer[],
+ stddev_r decimal[],
+ count_w integer[],
+ avg_w decimal[],
+ min_w integer[],
+ max_w integer[],
+ stddev_w decimal[]
+);
+
+create type long64_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ avg_r decimal[],
+ min_r bigint[],
+ max_r bigint[],
+ stddev_r decimal[],
+ count_w integer[],
+ avg_w decimal[],
+ min_w bigint[],
+ max_w bigint[],
+ stddev_w decimal[]
+);
+
+create type short_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ avg_r decimal[],
+ min_r smallint[],
+ max_r smallint[],
+ stddev_r decimal[],
+ count_w integer[],
+ avg_w decimal[],
+ min_w smallint[],
+ max_w smallint[],
+ stddev_w decimal[]
+);
+
+create type ulong_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ avg_r decimal[],
+ min_r ulong[],
+ max_r ulong[],
+ stddev_r decimal[],
+ count_w integer[],
+ avg_w decimal[],
+ min_w ulong[],
+ max_w ulong[],
+ stddev_w decimal[]
+);
+
+create type ulong64_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ avg_r decimal[],
+ min_r ulong64[],
+ max_r ulong64[],
+ stddev_r decimal[],
+ count_w integer[],
+ avg_w decimal[],
+ min_w ulong64[],
+ max_w ulong64[],
+ stddev_w decimal[]
+);
+
+create type ushort_array_agg_state as (
+ count integer,
+ count_errors integer,
+ count_r integer[],
+ avg_r decimal[],
+ min_r ushort[],
+ max_r ushort[],
+ stddev_r decimal[],
+ count_w integer[],
+ avg_w decimal[],
+ min_w ushort[],
+ max_w ushort[],
+ stddev_w decimal[]
+);
+
+-- Function to combine to aggregate state into a new one
+-- needed for the aggregate function to be used for partial aggregation
+CREATE OR REPLACE FUNCTION fn_double_combine(double_array_agg_state, double_array_agg_state)
+ RETURNS double_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result double_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::float8[], ARRAY[]::float8[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::float8[], ARRAY[]::float8[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.count_nan_r) AS nan_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.count_nan_w) AS nan_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.count_nan_r) AS nan_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.count_nan_w) AS nan_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2), array_agg(count_nan_r1+count_nan_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2), array_agg(count_nan_w1+count_nan_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_float_combine(float_array_agg_state, float_array_agg_state)
+ RETURNS float_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result float_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::real[], ARRAY[]::real[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::real[], ARRAY[]::real[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.count_nan_r) AS nan_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.count_nan_w) AS nan_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.count_nan_r) AS nan_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.count_nan_w) AS nan_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2), array_agg(count_nan_r1+count_nan_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2), array_agg(count_nan_w1+count_nan_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_long_combine(long_array_agg_state, long_array_agg_state)
+ RETURNS long_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result long_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.count_nan_r) AS nan_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.count_nan_w) AS nan_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.count_nan_r) AS nan_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.count_nan_w) AS nan_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_long64_combine(long64_array_agg_state, long64_array_agg_state)
+ RETURNS long64_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result long64_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::bigint[], ARRAY[]::bigint[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::bigint[], ARRAY[]::bigint[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_short_combine(short_array_agg_state, short_array_agg_state)
+ RETURNS short_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result short_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::smallint[], ARRAY[]::smallint[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::smallint[], ARRAY[]::smallint[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ulong_combine(ulong_array_agg_state, ulong_array_agg_state)
+ RETURNS ulong_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result ulong_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong[], ARRAY[]::ulong[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong[], ARRAY[]::ulong[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ulong64_combine(ulong64_array_agg_state, ulong64_array_agg_state)
+ RETURNS ulong64_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result ulong64_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong64[], ARRAY[]::ulong64[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong64[], ARRAY[]::ulong64[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ushort_combine(ushort_array_agg_state, ushort_array_agg_state)
+ RETURNS ushort_array_agg_state AS $$
+
+DECLARE
+ state1 ALIAS FOR $1;
+ state2 ALIAS FOR $2;
+ count integer;
+ count_errors integer;
+ result ushort_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Limit cases.
+ IF state1 is NULL
+ THEN
+ return state2;
+ END IF;
+
+ IF state2 is NULL
+ THEN
+ return state1;
+ END IF;
+
+ -- if there is a discrepancy in the arrays sizes
+ IF CARDINALITY(state1.avg_r) != CARDINALITY(state2.avg_r) OR CARDINALITY(state1.avg_w) != CARDINALITY(state2.avg_w) THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ushort[], ARRAY[]::ushort[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ushort[], ARRAY[]::ushort[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state1.count + state2.count;
+ count_errors := state1.count_errors + state2.count_errors;
+
+ WITH arrays AS(
+ SELECT
+ UNNEST(state1.count_r) AS count_r1, UNNEST(state1.avg_r) AS avg_r1,
+ UNNEST(state1.min_r) AS min_r1, UNNEST(state1.max_r) AS max_r1, UNNEST(state1.stddev_r) AS stddev_r1,
+ UNNEST(state1.count_w) AS count_w1, UNNEST(state1.avg_w) AS avg_w1,
+ UNNEST(state1.min_w) AS min_w1, UNNEST(state1.max_w) AS max_w1, UNNEST(state1.stddev_w) AS stddev_w1,
+ UNNEST(state2.count_r) AS count_r2, UNNEST(state2.avg_r) AS avg_r2,
+ UNNEST(state2.min_r) AS min_r2, UNNEST(state2.max_r) AS max_r2, UNNEST(state2.stddev_r) AS stddev_r2,
+ UNNEST(state2.count_w) AS count_w2, UNNEST(state2.avg_w) AS avg_w2,
+ UNNEST(state2.min_w) AS min_w2, UNNEST(state2.max_w) AS max_w2, UNNEST(state2.stddev_w) AS stddev_w2
+ )
+ SELECT count, count_errors,
+ array_agg(count_r1+count_r2),
+ array_agg(avg_r1 + (count_r2/(count_r1+count_r2))*(avg_r2-avg_r1)), array_agg(LEAST(min_r1, min_r2)), array_agg(GREATEST(max_r1, max_r2)),
+ array_agg(stddev_r1 + (count_r2*count_r1/count_r1+count_r2)*power(avg_r2 - avg_r1, 2)),
+ array_agg(count_w1+count_w2),
+ array_agg(avg_w1 + (count_w2/(count_w1+count_w2))*(avg_w2-avg_w1)), array_agg(LEAST(min_w1, min_w2)), array_agg(GREATEST(max_w1, max_w2)),
+ array_agg(stddev_w1 + (count_w2*count_w1/count_w1+count_w2)*power(avg_w2 - avg_w1, 2))
+ INTO result FROM arrays;
+ END IF;
+
+ return result;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+-- Function to compute next aggregate from last state and current row
+CREATE OR REPLACE FUNCTION fn_double_array_agg(double_array_agg_state,new_row att_array_devdouble)
+ RETURNS double_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result double_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN 0
+ WHEN read='Infinity'::float8 THEN 0
+ WHEN read='-Infinity'::float8 THEN 0
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN 1
+ WHEN read='Infinity'::float8 THEN 1
+ WHEN read='-Infinity'::float8 THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN NULL
+ WHEN read='Infinity'::float8 THEN NULL
+ WHEN read='-Infinity'::float8 THEN NULL
+ ELSE read::decimal
+ END
+ ), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN NULL
+ WHEN read='Infinity'::float8 THEN NULL
+ WHEN read='-Infinity'::float8 THEN NULL
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN 0
+ WHEN write='Infinity'::float8 THEN 0
+ WHEN write='-Infinity'::float8 THEN 0
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN 1
+ WHEN write='Infinity'::float8 THEN 1
+ WHEN write='-Infinity'::float8 THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN NULL
+ WHEN write='Infinity'::float8 THEN NULL
+ WHEN write='-Infinity'::float8 THEN NULL
+ ELSE write::decimal
+ END
+ ), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN NULL
+ WHEN write='Infinity'::float8 THEN NULL
+ WHEN write='-Infinity'::float8 THEN NULL
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::float8[], ARRAY[]::float8[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::float8[], ARRAY[]::float8[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.count_nan_r) AS nan_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.count_nan_w) AS nan_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read='NaN'::float8 THEN count_r
+ WHEN read='Infinity'::float8 THEN count_r
+ WHEN read='-Infinity'::float8 THEN count_r
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN nan_r + 1
+ WHEN read='Infinity'::float8 THEN nan_r + 1
+ WHEN read='-Infinity'::float8 THEN nan_r + 1
+ ELSE nan_r
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN avg_r
+ WHEN read='Infinity'::float8 THEN avg_r
+ WHEN read='-Infinity'::float8 THEN avg_r
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read::decimal
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal::decimal
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN min_r
+ WHEN read='Infinity'::float8 THEN min_r
+ WHEN read='-Infinity'::float8 THEN min_r
+ ELSE LEAST(read, min_r)
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN max_r
+ WHEN read='Infinity'::float8 THEN max_r
+ WHEN read='-Infinity'::float8 THEN max_r
+ ELSE GREATEST(read, max_r)
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN stddev_r
+ WHEN read='Infinity'::float8 THEN stddev_r
+ WHEN read='-Infinity'::float8 THEN stddev_r
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN count_w
+ WHEN write='Infinity'::float8 THEN count_w
+ WHEN write='-Infinity'::float8 THEN count_w
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN nan_w + 1
+ WHEN write='Infinity'::float8 THEN nan_w + 1
+ WHEN write='-Infinity'::float8 THEN nan_w + 1
+ ELSE nan_w
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN avg_w
+ WHEN write='Infinity'::float8 THEN avg_w
+ WHEN write='-Infinity'::float8 THEN avg_w
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN min_w
+ WHEN write='Infinity'::float8 THEN min_w
+ WHEN write='-Infinity'::float8 THEN min_w
+ ELSE LEAST(write, min_w)
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN max_w
+ WHEN write='Infinity'::float8 THEN max_w
+ WHEN write='-Infinity'::float8 THEN max_w
+ ELSE GREATEST(write, max_w)
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN stddev_w
+ WHEN write='Infinity'::float8 THEN stddev_w
+ WHEN write='-Infinity'::float8 THEN stddev_w
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+/*
+* Different method using compute_element_agg
+
+ SELECT n_count_r, n_count_nan_r, n_avg_r, n_min_r, n_max_r, n_stddev_r
+ , n_count_w, n_count_nan_w, n_avg_w, n_min_w, n_max_w, n_stddev_w
+ FROM compute_element_agg(
+ ( SELECT array_agg(ROW(read, write
+ , count_r, nan_r, avg_r, min_r, max_r, stddev_r
+ , count_w, nan_w, avg_w, min_w, max_w, stddev_w)::double_agg_input) from arrays )
+ ) as (n_count_r integer, n_count_nan_r integer, n_avg_r decimal, n_min_r double precision, n_max_r double precision, n_stddev_r decimal
+ , n_count_w integer, n_count_nan_w integer, n_avg_w decimal, n_min_w double precision, n_max_w double precision, n_stddev_w decimal)
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(aggregates.n_count_r), array_agg(aggregates.n_count_nan_r), array_agg(aggregates.n_avg_r), array_agg(aggregates.n_min_r), array_agg(aggregates.n_max_r), array_agg(aggregates.n_stddev_r)
+ , array_agg(aggregates.n_count_w), array_agg(aggregates.n_count_nan_w), array_agg(aggregates.n_avg_w), array_agg(aggregates.n_min_w), array_agg(aggregates.n_max_w), array_agg(aggregates.n_stddev_w)
+ into result from aggregates;
+*/
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_float_array_agg(float_array_agg_state,new_row att_array_devfloat)
+ RETURNS float_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result float_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN 0
+ WHEN read='Infinity'::float8 THEN 0
+ WHEN read='-Infinity'::float8 THEN 0
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN 1
+ WHEN read='Infinity'::float8 THEN 1
+ WHEN read='-Infinity'::float8 THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN NULL
+ WHEN read='Infinity'::float8 THEN NULL
+ WHEN read='-Infinity'::float8 THEN NULL
+ ELSE read::decimal
+ END
+ ), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read='NaN'::float8 THEN NULL
+ WHEN read='Infinity'::float8 THEN NULL
+ WHEN read='-Infinity'::float8 THEN NULL
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN 0
+ WHEN write='Infinity'::float8 THEN 0
+ WHEN write='-Infinity'::float8 THEN 0
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN 1
+ WHEN write='Infinity'::float8 THEN 1
+ WHEN write='-Infinity'::float8 THEN 1
+ ELSE 0
+ END
+ ), array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN NULL
+ WHEN write='Infinity'::float8 THEN NULL
+ WHEN write='-Infinity'::float8 THEN NULL
+ ELSE write::decimal
+ END
+ ), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write='NaN'::float8 THEN NULL
+ WHEN write='Infinity'::float8 THEN NULL
+ WHEN write='-Infinity'::float8 THEN NULL
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::real[], ARRAY[]::real[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::real[], ARRAY[]::real[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.count_nan_r) AS nan_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.count_nan_w) AS nan_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read='NaN'::float8 THEN count_r
+ WHEN read='Infinity'::float8 THEN count_r
+ WHEN read='-Infinity'::float8 THEN count_r
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN nan_r + 1
+ WHEN read='Infinity'::float8 THEN nan_r + 1
+ WHEN read='-Infinity'::float8 THEN nan_r + 1
+ ELSE nan_r
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN avg_r
+ WHEN read='Infinity'::float8 THEN avg_r
+ WHEN read='-Infinity'::float8 THEN avg_r
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN min_r
+ WHEN read='Infinity'::float8 THEN min_r
+ WHEN read='-Infinity'::float8 THEN min_r
+ ELSE LEAST(read, min_r)
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN max_r
+ WHEN read='Infinity'::float8 THEN max_r
+ WHEN read='-Infinity'::float8 THEN max_r
+ ELSE GREATEST(read, max_r)
+ END
+ ), array_agg(CASE
+ WHEN read='NaN'::float8 THEN stddev_r
+ WHEN read='Infinity'::float8 THEN stddev_r
+ WHEN read='-Infinity'::float8 THEN stddev_r
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN count_w
+ WHEN write='Infinity'::float8 THEN count_w
+ WHEN write='-Infinity'::float8 THEN count_w
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN nan_w + 1
+ WHEN write='Infinity'::float8 THEN nan_w + 1
+ WHEN write='-Infinity'::float8 THEN nan_w + 1
+ ELSE nan_w
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN avg_w
+ WHEN write='Infinity'::float8 THEN avg_w
+ WHEN write='-Infinity'::float8 THEN avg_w
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN min_w
+ WHEN write='Infinity'::float8 THEN min_w
+ WHEN write='-Infinity'::float8 THEN min_w
+ ELSE LEAST(write, min_w)
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN max_w
+ WHEN write='Infinity'::float8 THEN max_w
+ WHEN write='-Infinity'::float8 THEN max_w
+ ELSE GREATEST(write, max_w)
+ END
+ ), array_agg(CASE
+ WHEN write='NaN'::float8 THEN stddev_w
+ WHEN write='Infinity'::float8 THEN stddev_w
+ WHEN write='-Infinity'::float8 THEN stddev_w
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_long_array_agg(long_array_agg_state,new_row att_array_devlong)
+ RETURNS long_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result long_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(read::decimal), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(write::decimal), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::integer[], ARRAY[]::integer[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(LEAST(read, min_r)), array_agg(GREATEST(read, max_r))
+ , array_agg(CASE
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(LEAST(write, min_w)), array_agg(GREATEST(write, max_w))
+ , array_agg(CASE
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_long64_array_agg(long64_array_agg_state,new_row att_array_devlong64)
+ RETURNS long64_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result long64_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(read::decimal), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(write::decimal), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::bigint[], ARRAY[]::bigint[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::bigint[], ARRAY[]::bigint[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(LEAST(read, min_r)), array_agg(GREATEST(read, max_r))
+ , array_agg(CASE
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(LEAST(write, min_w)), array_agg(GREATEST(write, max_w))
+ , array_agg(CASE
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_short_array_agg(short_array_agg_state,new_row att_array_devshort)
+ RETURNS short_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result short_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(read::decimal), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(write::decimal), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::smallint[], ARRAY[]::smallint[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::smallint[], ARRAY[]::smallint[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(LEAST(read, min_r)), array_agg(GREATEST(read, max_r))
+ , array_agg(CASE
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(LEAST(write, min_w)), array_agg(GREATEST(write, max_w))
+ , array_agg(CASE
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ulong_array_agg(ulong_array_agg_state,new_row att_array_devulong)
+ RETURNS ulong_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result ulong_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(read::decimal), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(write::decimal), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong[], ARRAY[]::ulong[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong[], ARRAY[]::ulong[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(LEAST(read, min_r)), array_agg(GREATEST(read, max_r))
+ , array_agg(CASE
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(LEAST(write, min_w)), array_agg(GREATEST(write, max_w))
+ , array_agg(CASE
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ulong64_array_agg(ulong64_array_agg_state,new_row att_array_devulong64)
+ RETURNS ulong64_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result ulong64_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(read::decimal), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(write::decimal), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong64[], ARRAY[]::ulong64[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ulong64[], ARRAY[]::ulong64[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(LEAST(read, min_r)), array_agg(GREATEST(read, max_r))
+ , array_agg(CASE
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(LEAST(write, min_w)), array_agg(GREATEST(write, max_w))
+ , array_agg(CASE
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ushort_array_agg(ushort_array_agg_state,new_row att_array_devushort)
+ RETURNS ushort_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ count integer;
+ count_err integer;
+ result ushort_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ -- Increment error count if needed
+ IF new_row.att_error_desc_id > 0 THEN
+ count_err = 1;
+ ELSE
+ count_err = 0;
+ END IF;
+
+ IF state is NULL
+ THEN
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write)
+ SELECT 1, count_err,
+ array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(read::decimal), array_agg(read), array_agg(read), array_agg(
+ CASE
+ WHEN read IS NOT NULL THEN 0
+ ELSE read
+ END
+ ),
+ array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 1
+ ELSE 0
+ END
+ ), array_agg(write::decimal), array_agg(write), array_agg(write), array_agg(
+ CASE
+ WHEN write IS NOT NULL THEN 0
+ ELSE write
+ END
+ )
+ INTO result FROM arrays;
+ ELSE
+
+ IF CARDINALITY(state.avg_r) != CARDINALITY(new_row.value_r) or CARDINALITY(state.avg_w) != CARDINALITY(new_row.value_w)
+ THEN
+ SELECT 0, 0,
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ushort[], ARRAY[]::ushort[], ARRAY[]::decimal[],
+ ARRAY[]::integer[], ARRAY[]::decimal[], ARRAY[]::ushort[], ARRAY[]::ushort[], ARRAY[]::decimal[]
+ INTO result;
+ ELSE
+
+ count := state.count + 1;
+ WITH arrays AS(
+ SELECT UNNEST(new_row.value_r) AS read, UNNEST(new_row.value_w) AS write,
+ UNNEST(state.count_r) AS count_r, UNNEST(state.avg_r) AS avg_r,
+ UNNEST(state.min_r) AS min_r, UNNEST(state.max_r) AS max_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.avg_w) AS avg_w,
+ UNNEST(state.min_w) AS min_w, UNNEST(state.max_w) AS max_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT count, state.count_errors+count_err
+ , array_agg(CASE
+ WHEN read IS NOT NULL THEN count_r+1
+ ELSE count_r
+ END
+ ), array_agg(CASE
+ WHEN read IS NULL THEN avg_r
+ WHEN avg_r IS NULL THEN read
+ ELSE avg_r + (read::decimal-avg_r)/(count_r+1.)::decimal
+ END
+ ), array_agg(LEAST(read, min_r)), array_agg(GREATEST(read, max_r))
+ , array_agg(CASE
+ WHEN read IS NULL THEN stddev_r
+ WHEN stddev_r IS NULL THEN 0
+ ELSE stddev_r + ((count_r+0.)/(count_r+1.))::decimal*power(read::decimal - avg_r, 2)
+ END
+ ), array_agg(CASE
+ WHEN write IS NOT NULL THEN count_w+1
+ ELSE count_w
+ END
+ ), array_agg(CASE
+ WHEN write IS NULL THEN avg_w
+ WHEN avg_w IS NULL THEN write
+ ELSE avg_w + (write::decimal-avg_w)/(count_w+1.)::decimal
+ END
+ ), array_agg(LEAST(write, min_w)), array_agg(GREATEST(write, max_w))
+ , array_agg(CASE
+ WHEN write IS NULL THEN stddev_w
+ WHEN stddev_w IS NULL THEN 0
+ ELSE stddev_w + ((count_w+0.)/(count_w+1.))::decimal*power(write::decimal - avg_w, 2)
+ END
+ )
+ INTO result FROM arrays;
+ END IF;
+ END IF;
+
+ return result;
+
+END;
+$$
+LANGUAGE 'plpgsql';
+
+-- Function to compute the real aggregate results from the internal state
+-- in this case only the stddev has to be computed
+CREATE OR REPLACE FUNCTION fn_double_array_final(double_array_agg_state)
+ RETURNS double_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result double_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.count_nan_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.count_nan_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END
+ )
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_float_array_final(float_array_agg_state)
+ RETURNS float_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result float_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.count_nan_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.count_nan_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_long_array_final(long_array_agg_state)
+ RETURNS long_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result long_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_long64_array_final(long64_array_agg_state)
+ RETURNS long64_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result long64_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_short_array_final(short_array_agg_state)
+ RETURNS short_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result short_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ulong_array_final(ulong_array_agg_state)
+ RETURNS ulong_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result ulong_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ulong64_array_final(ulong64_array_agg_state)
+ RETURNS ulong64_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result ulong64_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+CREATE OR REPLACE FUNCTION fn_ushort_array_final(ushort_array_agg_state)
+ RETURNS ushort_array_agg_state AS $$
+
+DECLARE
+ state ALIAS FOR $1;
+ result ushort_array_agg_state%ROWTYPE;
+
+BEGIN
+
+ IF state IS NULL
+ THEN
+ return NULL;
+ END IF;
+
+ IF state.count = 0 THEN
+ return NULL;
+
+ ELSE
+ WITH arrays AS(
+ SELECT UNNEST(state.count_r) AS count_r, UNNEST(state.stddev_r) AS stddev_r,
+ UNNEST(state.count_w) AS count_w, UNNEST(state.stddev_w) AS stddev_w
+ )
+ SELECT state.count, state.count_errors,
+ state.count_r, state.avg_r,
+ state.min_r, state.max_r, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_r/(count_r))
+ END
+ ),
+ state.count_w, state.avg_w,
+ state.min_w, state.max_w, array_agg(CASE
+ WHEN count_r=0 THEN NULL
+ WHEN count_r=1 THEN 0
+ ELSE sqrt(stddev_w/(count_w))
+ END)
+ INTO result FROM arrays;
+
+ return result;
+
+ END IF;
+END;
+$$
+LANGUAGE 'plpgsql';
+
+-- Aggregate function declaration
+CREATE AGGREGATE double_array_aggregate(att_array_devdouble)
+(
+ sfunc = fn_double_array_agg,
+ stype = double_array_agg_state,
+ combinefunc = fn_double_combine,
+ finalfunc = fn_double_array_final
+);
+
+CREATE AGGREGATE float_array_aggregate(att_array_devfloat)
+(
+ sfunc = fn_float_array_agg,
+ stype = float_array_agg_state,
+ combinefunc = fn_float_combine,
+ finalfunc = fn_float_array_final
+);
+
+CREATE AGGREGATE long_array_aggregate(att_array_devlong)
+(
+ sfunc = fn_long_array_agg,
+ stype = long_array_agg_state,
+ combinefunc = fn_long_combine,
+ finalfunc = fn_long_array_final
+);
+
+CREATE AGGREGATE long64_array_aggregate(att_array_devlong64)
+(
+ sfunc = fn_long64_array_agg,
+ stype = long64_array_agg_state,
+ combinefunc = fn_long64_combine,
+ finalfunc = fn_long64_array_final
+);
+
+CREATE AGGREGATE short_array_aggregate(att_array_devshort)
+(
+ sfunc = fn_short_array_agg,
+ stype = short_array_agg_state,
+ combinefunc = fn_short_combine,
+ finalfunc = fn_short_array_final
+);
+
+CREATE AGGREGATE ulong_array_aggregate(att_array_devulong)
+(
+ sfunc = fn_ulong_array_agg,
+ stype = ulong_array_agg_state,
+ combinefunc = fn_ulong_combine,
+ finalfunc = fn_ulong_array_final
+);
+
+CREATE AGGREGATE ulong64_array_aggregate(att_array_devulong64)
+(
+ sfunc = fn_ulong64_array_agg,
+ stype = ulong64_array_agg_state,
+ combinefunc = fn_ulong64_combine,
+ finalfunc = fn_ulong64_array_final
+);
+
+CREATE AGGREGATE ushort_array_aggregate(att_array_devushort)
+(
+ sfunc = fn_ushort_array_agg,
+ stype = ushort_array_agg_state,
+ combinefunc = fn_ushort_combine,
+ finalfunc = fn_ushort_array_final
+);
diff --git a/docker-compose/timescaledb/resources/07_hdb_ext_arrays_aggregates.sql b/docker-compose/timescaledb/resources/07_hdb_ext_arrays_aggregates.sql
new file mode 100644
index 000000000..028712989
--- /dev/null
+++ b/docker-compose/timescaledb/resources/07_hdb_ext_arrays_aggregates.sql
@@ -0,0 +1,328 @@
+-- -----------------------------------------------------------------------------
+-- This file is part of the hdbpp-timescale-project
+--
+-- Copyright (C) : 2014-2019
+-- European Synchrotron Radiation Facility
+-- BP 220, Grenoble 38043, FRANCE
+--
+-- libhdb++timescale is free software: you can redistribute it and/or modify
+-- it under the terms of the Lesser GNU General Public License as published by
+-- the Free Software Foundation, either version 3 of the License, or
+-- (at your option) any later version.
+--
+-- libhdb++timescale is distributed in the hope that it will be useful,
+-- but WITHOUT ANY WARRANTY; without even the implied warranty of
+-- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Lesser
+-- GNU General Public License for more details.
+--
+-- You should have received a copy of the Lesser GNU General Public License
+-- along with libhdb++timescale. If not, see <http://www.gnu.org/licenses/>.
+-- -----------------------------------------------------------------------------
+
+-- Continuous aggregates views for the array attributes.
+\c hdb
+-- Double attributes
+CREATE MATERIALIZED VIEW cagg_array_devdouble_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, count_nan_r, mean_r, min_r, max_r, stddev_r
+ , count_w, count_nan_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (double_array_aggregate(t)).count, (double_array_aggregate(t)).count_errors
+ , (double_array_aggregate(t)).count_r, (double_array_aggregate(t)).count_nan_r, (double_array_aggregate(t)).avg_r::float8[], (double_array_aggregate(t)).min_r, (double_array_aggregate(t)).max_r, (double_array_aggregate(t)).stddev_r::float8[]
+ , (double_array_aggregate(t)).count_w, (double_array_aggregate(t)).count_nan_w, (double_array_aggregate(t)).avg_w::float8[], (double_array_aggregate(t)).min_w, (double_array_aggregate(t)).max_w, (double_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devdouble as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devdouble_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, count_nan_r, mean_r, min_r, max_r, stddev_r
+ , count_w, count_nan_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (double_array_aggregate(t)).count, (double_array_aggregate(t)).count_errors
+ , (double_array_aggregate(t)).count_r, (double_array_aggregate(t)).count_nan_r, (double_array_aggregate(t)).avg_r::float8[], (double_array_aggregate(t)).min_r, (double_array_aggregate(t)).max_r, (double_array_aggregate(t)).stddev_r::float8[]
+ , (double_array_aggregate(t)).count_w, (double_array_aggregate(t)).count_nan_w, (double_array_aggregate(t)).avg_w::float8[], (double_array_aggregate(t)).min_w, (double_array_aggregate(t)).max_w, (double_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devdouble as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devdouble_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, count_nan_r, mean_r, min_r, max_r, stddev_r
+ , count_w, count_nan_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (double_array_aggregate(t)).count, (double_array_aggregate(t)).count_errors
+ , (double_array_aggregate(t)).count_r, (double_array_aggregate(t)).count_nan_r, (double_array_aggregate(t)).avg_r::float8[], (double_array_aggregate(t)).min_r, (double_array_aggregate(t)).max_r, (double_array_aggregate(t)).stddev_r::float8[]
+ , (double_array_aggregate(t)).count_w, (double_array_aggregate(t)).count_nan_w, (double_array_aggregate(t)).avg_w::float8[], (double_array_aggregate(t)).min_w, (double_array_aggregate(t)).max_w, (double_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devdouble as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Float attributes
+CREATE MATERIALIZED VIEW cagg_array_devfloat_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, count_nan_r, mean_r, min_r, max_r, stddev_r
+ , count_w, count_nan_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (float_array_aggregate(t)).count, (float_array_aggregate(t)).count_errors
+ , (float_array_aggregate(t)).count_r, (float_array_aggregate(t)).count_nan_r, (float_array_aggregate(t)).avg_r::float8[], (float_array_aggregate(t)).min_r, (float_array_aggregate(t)).max_r, (float_array_aggregate(t)).stddev_r::float8[]
+ , (float_array_aggregate(t)).count_w, (float_array_aggregate(t)).count_nan_w, (float_array_aggregate(t)).avg_w::float8[], (float_array_aggregate(t)).min_w, (float_array_aggregate(t)).max_w, (float_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devfloat as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devfloat_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, count_nan_r, mean_r, min_r, max_r, stddev_r
+ , count_w, count_nan_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (float_array_aggregate(t)).count, (float_array_aggregate(t)).count_errors
+ , (float_array_aggregate(t)).count_r, (float_array_aggregate(t)).count_nan_r, (float_array_aggregate(t)).avg_r::float8[], (float_array_aggregate(t)).min_r, (float_array_aggregate(t)).max_r, (float_array_aggregate(t)).stddev_r::float8[]
+ , (float_array_aggregate(t)).count_w, (float_array_aggregate(t)).count_nan_w, (float_array_aggregate(t)).avg_w::float8[], (float_array_aggregate(t)).min_w, (float_array_aggregate(t)).max_w, (float_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devfloat as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devfloat_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, count_nan_r, mean_r, min_r, max_r, stddev_r
+ , count_w, count_nan_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (float_array_aggregate(t)).count, (float_array_aggregate(t)).count_errors
+ , (float_array_aggregate(t)).count_r, (float_array_aggregate(t)).count_nan_r, (float_array_aggregate(t)).avg_r::float8[], (float_array_aggregate(t)).min_r, (float_array_aggregate(t)).max_r, (float_array_aggregate(t)).stddev_r::float8[]
+ , (float_array_aggregate(t)).count_w, (float_array_aggregate(t)).count_nan_w, (float_array_aggregate(t)).avg_w::float8[], (float_array_aggregate(t)).min_w, (float_array_aggregate(t)).max_w, (float_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devfloat as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Long attributes
+CREATE MATERIALIZED VIEW cagg_array_devlong_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (long_array_aggregate(t)).count, (long_array_aggregate(t)).count_errors
+ , (long_array_aggregate(t)).count_r, (long_array_aggregate(t)).avg_r::float8[], (long_array_aggregate(t)).min_r, (long_array_aggregate(t)).max_r, (long_array_aggregate(t)).stddev_r::float8[]
+ , (long_array_aggregate(t)).count_w, (long_array_aggregate(t)).avg_w::float8[], (long_array_aggregate(t)).min_w, (long_array_aggregate(t)).max_w, (long_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devlong as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devlong_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (long_array_aggregate(t)).count, (long_array_aggregate(t)).count_errors
+ , (long_array_aggregate(t)).count_r, (long_array_aggregate(t)).avg_r::float8[], (long_array_aggregate(t)).min_r, (long_array_aggregate(t)).max_r, (long_array_aggregate(t)).stddev_r::float8[]
+ , (long_array_aggregate(t)).count_w, (long_array_aggregate(t)).avg_w::float8[], (long_array_aggregate(t)).min_w, (long_array_aggregate(t)).max_w, (long_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devlong as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devlong_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (long_array_aggregate(t)).count, (long_array_aggregate(t)).count_errors
+ , (long_array_aggregate(t)).count_r, (long_array_aggregate(t)).avg_r::float8[], (long_array_aggregate(t)).min_r, (long_array_aggregate(t)).max_r, (long_array_aggregate(t)).stddev_r::float8[]
+ , (long_array_aggregate(t)).count_w, (long_array_aggregate(t)).avg_w::float8[], (long_array_aggregate(t)).min_w, (long_array_aggregate(t)).max_w, (long_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devlong as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Long 64 attributes
+CREATE MATERIALIZED VIEW cagg_array_devlong64_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (long64_array_aggregate(t)).count, (long64_array_aggregate(t)).count_errors
+ , (long64_array_aggregate(t)).count_r, (long64_array_aggregate(t)).avg_r::float8[], (long64_array_aggregate(t)).min_r, (long64_array_aggregate(t)).max_r, (long64_array_aggregate(t)).stddev_r::float8[]
+ , (long64_array_aggregate(t)).count_w, (long64_array_aggregate(t)).avg_w::float8[], (long64_array_aggregate(t)).min_w, (long64_array_aggregate(t)).max_w, (long64_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devlong64 as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devlong64_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (long64_array_aggregate(t)).count, (long64_array_aggregate(t)).count_errors
+ , (long64_array_aggregate(t)).count_r, (long64_array_aggregate(t)).avg_r::float8[], (long64_array_aggregate(t)).min_r, (long64_array_aggregate(t)).max_r, (long64_array_aggregate(t)).stddev_r::float8[]
+ , (long64_array_aggregate(t)).count_w, (long64_array_aggregate(t)).avg_w::float8[], (long64_array_aggregate(t)).min_w, (long64_array_aggregate(t)).max_w, (long64_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devlong64 as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devlong64_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (long64_array_aggregate(t)).count, (long64_array_aggregate(t)).count_errors
+ , (long64_array_aggregate(t)).count_r, (long64_array_aggregate(t)).avg_r::float8[], (long64_array_aggregate(t)).min_r, (long64_array_aggregate(t)).max_r, (long64_array_aggregate(t)).stddev_r::float8[]
+ , (long64_array_aggregate(t)).count_w, (long64_array_aggregate(t)).avg_w::float8[], (long64_array_aggregate(t)).min_w, (long64_array_aggregate(t)).max_w, (long64_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devlong64 as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Short attributes
+CREATE MATERIALIZED VIEW cagg_array_devshort_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (short_array_aggregate(t)).count, (short_array_aggregate(t)).count_errors
+ , (short_array_aggregate(t)).count_r, (short_array_aggregate(t)).avg_r::float8[], (short_array_aggregate(t)).min_r, (short_array_aggregate(t)).max_r, (short_array_aggregate(t)).stddev_r::float8[]
+ , (short_array_aggregate(t)).count_w, (short_array_aggregate(t)).avg_w::float8[], (short_array_aggregate(t)).min_w, (short_array_aggregate(t)).max_w, (short_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devshort as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devshort_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (short_array_aggregate(t)).count, (short_array_aggregate(t)).count_errors
+ , (short_array_aggregate(t)).count_r, (short_array_aggregate(t)).avg_r::float8[], (short_array_aggregate(t)).min_r, (short_array_aggregate(t)).max_r, (short_array_aggregate(t)).stddev_r::float8[]
+ , (short_array_aggregate(t)).count_w, (short_array_aggregate(t)).avg_w::float8[], (short_array_aggregate(t)).min_w, (short_array_aggregate(t)).max_w, (short_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devshort as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devshort_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (short_array_aggregate(t)).count, (short_array_aggregate(t)).count_errors
+ , (short_array_aggregate(t)).count_r, (short_array_aggregate(t)).avg_r::float8[], (short_array_aggregate(t)).min_r, (short_array_aggregate(t)).max_r, (short_array_aggregate(t)).stddev_r::float8[]
+ , (short_array_aggregate(t)).count_w, (short_array_aggregate(t)).avg_w::float8[], (short_array_aggregate(t)).min_w, (short_array_aggregate(t)).max_w, (short_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devshort as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Unsigned long attributes
+CREATE MATERIALIZED VIEW cagg_array_devulong_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (ulong_array_aggregate(t)).count, (ulong_array_aggregate(t)).count_errors
+ , (ulong_array_aggregate(t)).count_r, (ulong_array_aggregate(t)).avg_r::float8[], (ulong_array_aggregate(t)).min_r, (ulong_array_aggregate(t)).max_r, (ulong_array_aggregate(t)).stddev_r::float8[]
+ , (ulong_array_aggregate(t)).count_w, (ulong_array_aggregate(t)).avg_w::float8[], (ulong_array_aggregate(t)).min_w, (ulong_array_aggregate(t)).max_w, (ulong_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devulong as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devulong_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (ulong_array_aggregate(t)).count, (ulong_array_aggregate(t)).count_errors
+ , (ulong_array_aggregate(t)).count_r, (ulong_array_aggregate(t)).avg_r::float8[], (ulong_array_aggregate(t)).min_r, (ulong_array_aggregate(t)).max_r, (ulong_array_aggregate(t)).stddev_r::float8[]
+ , (ulong_array_aggregate(t)).count_w, (ulong_array_aggregate(t)).avg_w::float8[], (ulong_array_aggregate(t)).min_w, (ulong_array_aggregate(t)).max_w, (ulong_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devulong as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devulong_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (ulong_array_aggregate(t)).count, (ulong_array_aggregate(t)).count_errors
+ , (ulong_array_aggregate(t)).count_r, (ulong_array_aggregate(t)).avg_r::float8[], (ulong_array_aggregate(t)).min_r, (ulong_array_aggregate(t)).max_r, (ulong_array_aggregate(t)).stddev_r::float8[]
+ , (ulong_array_aggregate(t)).count_w, (ulong_array_aggregate(t)).avg_w::float8[], (ulong_array_aggregate(t)).min_w, (ulong_array_aggregate(t)).max_w, (ulong_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devulong as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Unsigned long 64 attributes
+CREATE MATERIALIZED VIEW cagg_array_devulong64_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (ulong64_array_aggregate(t)).count, (ulong64_array_aggregate(t)).count_errors
+ , (ulong64_array_aggregate(t)).count_r, (ulong64_array_aggregate(t)).avg_r::float8[], (ulong64_array_aggregate(t)).min_r, (ulong64_array_aggregate(t)).max_r, (ulong64_array_aggregate(t)).stddev_r::float8[]
+ , (ulong64_array_aggregate(t)).count_w, (ulong64_array_aggregate(t)).avg_w::float8[], (ulong64_array_aggregate(t)).min_w, (ulong64_array_aggregate(t)).max_w, (ulong64_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devulong64 as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devulong64_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (ulong64_array_aggregate(t)).count, (ulong64_array_aggregate(t)).count_errors
+ , (ulong64_array_aggregate(t)).count_r, (ulong64_array_aggregate(t)).avg_r::float8[], (ulong64_array_aggregate(t)).min_r, (ulong64_array_aggregate(t)).max_r, (ulong64_array_aggregate(t)).stddev_r::float8[]
+ , (ulong64_array_aggregate(t)).count_w, (ulong64_array_aggregate(t)).avg_w::float8[], (ulong64_array_aggregate(t)).min_w, (ulong64_array_aggregate(t)).max_w, (ulong64_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devulong64 as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devulong64_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (ulong64_array_aggregate(t)).count, (ulong64_array_aggregate(t)).count_errors
+ , (ulong64_array_aggregate(t)).count_r, (ulong64_array_aggregate(t)).avg_r::float8[], (ulong64_array_aggregate(t)).min_r, (ulong64_array_aggregate(t)).max_r, (ulong64_array_aggregate(t)).stddev_r::float8[]
+ , (ulong64_array_aggregate(t)).count_w, (ulong64_array_aggregate(t)).avg_w::float8[], (ulong64_array_aggregate(t)).min_w, (ulong64_array_aggregate(t)).max_w, (ulong64_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devulong64 as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Unsigned short attributes
+CREATE MATERIALIZED VIEW cagg_array_devushort_1hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 hour', data_time), (ushort_array_aggregate(t)).count, (ushort_array_aggregate(t)).count_errors
+ , (ushort_array_aggregate(t)).count_r, (ushort_array_aggregate(t)).avg_r::float8[], (ushort_array_aggregate(t)).min_r, (ushort_array_aggregate(t)).max_r, (ushort_array_aggregate(t)).stddev_r::float8[]
+ , (ushort_array_aggregate(t)).count_w, (ushort_array_aggregate(t)).avg_w::float8[], (ushort_array_aggregate(t)).min_w, (ushort_array_aggregate(t)).max_w, (ushort_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devushort as t
+ GROUP BY time_bucket('1 hour', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devushort_8hour(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('8 hours', data_time), (ushort_array_aggregate(t)).count, (ushort_array_aggregate(t)).count_errors
+ , (ushort_array_aggregate(t)).count_r, (ushort_array_aggregate(t)).avg_r::float8[], (ushort_array_aggregate(t)).min_r, (ushort_array_aggregate(t)).max_r, (ushort_array_aggregate(t)).stddev_r::float8[]
+ , (ushort_array_aggregate(t)).count_w, (ushort_array_aggregate(t)).avg_w::float8[], (ushort_array_aggregate(t)).min_w, (ushort_array_aggregate(t)).max_w, (ushort_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devushort as t
+ GROUP BY time_bucket('8 hours', data_time), att_conf_id;
+
+CREATE MATERIALIZED VIEW cagg_array_devushort_1day(
+ att_conf_id, data_time, count_rows, count_errors
+ , count_r, mean_r, min_r, max_r, stddev_r
+ , count_w, mean_w, min_w, max_w, stddev_w
+ ) WITH (timescaledb.continuous)
+ AS SELECT att_conf_id, time_bucket('1 day', data_time), (ushort_array_aggregate(t)).count, (ushort_array_aggregate(t)).count_errors
+ , (ushort_array_aggregate(t)).count_r, (ushort_array_aggregate(t)).avg_r::float8[], (ushort_array_aggregate(t)).min_r, (ushort_array_aggregate(t)).max_r, (ushort_array_aggregate(t)).stddev_r::float8[]
+ , (ushort_array_aggregate(t)).count_w, (ushort_array_aggregate(t)).avg_w::float8[], (ushort_array_aggregate(t)).min_w, (ushort_array_aggregate(t)).max_w, (ushort_array_aggregate(t)).stddev_w::float8[]
+ FROM att_array_devushort as t
+ GROUP BY time_bucket('1 day', data_time), att_conf_id;
+
+-- Drop all the views
+-- DROP VIEW cagg_array_devdouble_1hour CASCADE;
+-- DROP VIEW cagg_array_devdouble_8hour CASCADE;
+-- DROP VIEW cagg_array_devdouble_1day CASCADE;
+
+-- DROP VIEW cagg_array_devfloat_1hour CASCADE;
+-- DROP VIEW cagg_array_devfloat_8hour CASCADE;
+-- DROP VIEW cagg_array_devfloat_1day CASCADE;
+
+-- DROP VIEW cagg_array_devlong_1hour CASCADE;
+-- DROP VIEW cagg_array_devlong_8hour CASCADE;
+-- DROP VIEW cagg_array_devlong_1day CASCADE;
+
+-- DROP VIEW cagg_array_devlong64_1hour CASCADE;
+-- DROP VIEW cagg_array_devlong64_8hour CASCADE;
+-- DROP VIEW cagg_array_devlong64_1day CASCADE;
+
+-- DROP VIEW cagg_array_devshort_1hour CASCADE;
+-- DROP VIEW cagg_array_devshort_8hour CASCADE;
+-- DROP VIEW cagg_array_devshort_1day CASCADE;
+
+-- DROP VIEW cagg_array_devulong_1hour CASCADE;
+-- DROP VIEW cagg_array_devulong_8hour CASCADE;
+-- DROP VIEW cagg_array_devulong_1day CASCADE;
+
+-- DROP VIEW cagg_array_devulong64_1hour CASCADE;
+-- DROP VIEW cagg_array_devulong64_8hour CASCADE;
+-- DROP VIEW cagg_array_devulong64_1day CASCADE;
+
+-- DROP VIEW cagg_array_devushort_1hour CASCADE;
+-- DROP VIEW cagg_array_devushort_8hour CASCADE;
+-- DROP VIEW cagg_array_devushort_1day CASCADE;
+
diff --git a/docker-compose/timescaledb/resources/08_hdb_ext_compress_policy.sql b/docker-compose/timescaledb/resources/08_hdb_ext_compress_policy.sql
new file mode 100644
index 000000000..a815950e8
--- /dev/null
+++ b/docker-compose/timescaledb/resources/08_hdb_ext_compress_policy.sql
@@ -0,0 +1,118 @@
+-- -----------------------------------------------------------------------------
+-- This file is part of the hdbpp-timescale-project
+--
+-- Copyright (C) : 2014-2019
+-- European Synchrotron Radiation Facility
+-- BP 220, Grenoble 38043, FRANCE
+--
+-- libhdb++timescale is free software: you can redistribute it and/or modify
+-- it under the terms of the Lesser GNU General Public License as published by
+-- the Free Software Foundation, either version 3 of the License, or
+-- (at your option) any later version.
+--
+-- libhdb++timescale is distributed in the hope that it will be useful,
+-- but WITHOUT ANY WARRANTY; without even the implied warranty of
+-- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Lesser
+-- GNU General Public License for more details.
+--
+-- You should have received a copy of the Lesser GNU General Public License
+-- along with libhdb++timescale. If not, see <http://www.gnu.org/licenses/>.
+-- -----------------------------------------------------------------------------
+\c hdb
+-- Compress chunk policy
+-- Allow compression on the table
+ALTER TABLE att_scalar_devboolean SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devdouble SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devfloat SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devencoded SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devenum SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devstate SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devstring SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devuchar SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devulong SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devulong64 SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devlong64 SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devlong SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devushort SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_scalar_devshort SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+
+ALTER TABLE att_array_devboolean SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devdouble SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devfloat SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devencoded SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devenum SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devstate SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devstring SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devuchar SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devulong SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devulong64 SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devlong64 SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devlong SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devushort SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+ALTER TABLE att_array_devshort SET(timescaledb.compress, timescaledb.compress_segmentby = 'att_conf_id, att_error_desc_id', timescaledb.compress_orderby = 'data_time DESC');
+
+DO $$ BEGIN
+ IF (SELECT extversion>'2.0.0' FROM pg_extension where extname = 'timescaledb') THEN
+ -- If using timescaledb v2
+ PERFORM add_compression_policy('att_scalar_devboolean', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devdouble', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devfloat', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devencoded', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devenum', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devstate', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devstring', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devuchar', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devulong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devulong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devlong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devlong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devushort', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_scalar_devshort', INTERVAL '200d', if_not_exists => true);
+
+ PERFORM add_compression_policy('att_array_devboolean', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devdouble', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devfloat', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devencoded', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devenum', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devstate', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devstring', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devuchar', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devulong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devulong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devlong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devlong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devushort', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compression_policy('att_array_devshort', INTERVAL '200d', if_not_exists => true);
+ ELSE
+ -- If using timescaledb v1.7
+ PERFORM add_compress_chunks_policy('att_scalar_devboolean', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devdouble', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devfloat', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devencoded', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devenum', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devstate', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devstring', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devuchar', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devulong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devulong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devlong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devlong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devushort', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_scalar_devshort', INTERVAL '200d', if_not_exists => true);
+
+ PERFORM add_compress_chunks_policy('att_array_devboolean', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devdouble', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devfloat', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devencoded', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devenum', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devstate', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devstring', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devuchar', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devulong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devulong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devlong64', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devlong', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devushort', INTERVAL '200d', if_not_exists => true);
+ PERFORM add_compress_chunks_policy('att_array_devshort', INTERVAL '200d', if_not_exists => true);
+ END IF;
+END $$;
diff --git a/docker-compose/timescaledb/resources/09_hdb_ext_import.sql b/docker-compose/timescaledb/resources/09_hdb_ext_import.sql
new file mode 100644
index 000000000..f08cc3610
--- /dev/null
+++ b/docker-compose/timescaledb/resources/09_hdb_ext_import.sql
@@ -0,0 +1,19 @@
+\c hdb
+
+CREATE OR REPLACE FUNCTION expand_name() RETURNS TRIGGER AS $$
+DECLARE
+ len integer;
+BEGIN
+ IF (NEW.cs_name <> '' AND NEW.domain <> '' AND NEW.family <> '' AND NEW.member <> '' AND NEW.name <> '') IS NOT TRUE THEN
+ len = (SELECT cardinality((SELECT regexp_split_to_array(NEW.att_name, E'/'))));
+ NEW.name := (SELECT split_part(NEW.att_name, '/', len));
+ NEW.member := (SELECT split_part(NEW.att_name, '/', len - 1));
+ NEW.family := (SELECT split_part(NEW.att_name, '/', len - 2));
+ NEW.domain := (SELECT split_part(NEW.att_name, '/', len - 3));
+ NEW.cs_name := (SELECT split_part(NEW.att_name, '/', len - 4));
+ END IF;
+ RETURN NEW;
+END
+$$ LANGUAGE plpgsql;
+
+CREATE TRIGGER expand_name_trigger BEFORE INSERT ON att_conf FOR EACH ROW EXECUTE PROCEDURE expand_name();
diff --git a/docker-compose/timescaledb/resources/10_hdb_ext_reorder_policy.sql b/docker-compose/timescaledb/resources/10_hdb_ext_reorder_policy.sql
new file mode 100644
index 000000000..e8e0f3911
--- /dev/null
+++ b/docker-compose/timescaledb/resources/10_hdb_ext_reorder_policy.sql
@@ -0,0 +1,53 @@
+-- -----------------------------------------------------------------------------
+-- This file is part of the hdbpp-timescale-project
+--
+-- Copyright (C) : 2014-2019
+-- European Synchrotron Radiation Facility
+-- BP 220, Grenoble 38043, FRANCE
+--
+-- libhdb++timescale is free software: you can redistribute it and/or modify
+-- it under the terms of the Lesser GNU General Public License as published by
+-- the Free Software Foundation, either version 3 of the License, or
+-- (at your option) any later version.
+--
+-- libhdb++timescale is distributed in the hope that it will be useful,
+-- but WITHOUT ANY WARRANTY; without even the implied warranty of
+-- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Lesser
+-- GNU General Public License for more details.
+--
+-- You should have received a copy of the Lesser GNU General Public License
+-- along with libhdb++timescale. If not, see <http://www.gnu.org/licenses/>.
+-- -----------------------------------------------------------------------------
+
+\c hdb
+-- Reorder chunk policy
+
+SELECT add_reorder_policy('att_scalar_devboolean', 'att_scalar_devboolean_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devdouble', 'att_scalar_devdouble_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devfloat', 'att_scalar_devfloat_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devencoded', 'att_scalar_devencoded_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devenum', 'att_scalar_devenum_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devstate', 'att_scalar_devstate_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devstring', 'att_scalar_devstring_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devuchar', 'att_scalar_devuchar_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devulong', 'att_scalar_devulong_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devulong64', 'att_scalar_devulong64_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devlong64', 'att_scalar_devlong64_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devlong', 'att_scalar_devlong_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devushort', 'att_scalar_devushort_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_scalar_devshort', 'att_scalar_devshort_att_conf_id_data_time_idx', if_not_exists => true);
+
+SELECT add_reorder_policy('att_array_devboolean', 'att_array_devboolean_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devdouble', 'att_array_devdouble_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devfloat', 'att_array_devfloat_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devencoded', 'att_array_devencoded_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devenum', 'att_array_devenum_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devstate', 'att_array_devstate_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devstring', 'att_array_devstring_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devuchar', 'att_array_devuchar_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devulong', 'att_array_devulong_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devulong64', 'att_array_devulong64_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devlong64', 'att_array_devlong64_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devlong', 'att_array_devlong_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devushort', 'att_array_devushort_att_conf_id_data_time_idx', if_not_exists => true);
+SELECT add_reorder_policy('att_array_devshort', 'att_array_devshort_att_conf_id_data_time_idx', if_not_exists => true);
diff --git a/docker-compose/timescaledb/resources/05_lofar_func.sh b/docker-compose/timescaledb/resources/11_lofar_func.sh
similarity index 100%
rename from docker-compose/timescaledb/resources/05_lofar_func.sh
rename to docker-compose/timescaledb/resources/11_lofar_func.sh
diff --git a/docker-compose/timescaledb/resources/06_lofar_views.sql b/docker-compose/timescaledb/resources/12_lofar_views.sql
similarity index 100%
rename from docker-compose/timescaledb/resources/06_lofar_views.sql
rename to docker-compose/timescaledb/resources/12_lofar_views.sql
diff --git a/docker-compose/timescaledb/resources/07_cleanup.sql b/docker-compose/timescaledb/resources/13_cleanup.sql
similarity index 100%
rename from docker-compose/timescaledb/resources/07_cleanup.sql
rename to docker-compose/timescaledb/resources/13_cleanup.sql
--
GitLab