Tests for nested queries and against aws by grajoria · Pull Request #72 · ceph/s3select

grajoria · 2021-03-10T08:34:16Z

Added tests to run nested complex queries which are combinations of
various functions. Equivalent queries runs for aws and ceph, then results are matched

Signed-off-by: Girjesh Rajoria grajoria@redhat.com

galsalomon66 · 2021-03-26T07:55:49Z

@@ -0,0 +1,4 @@
+select to_string(to_timestamp('2009-09-17T17:56:06.234567Z'), substring(' athmywopgss-nghjkl', 3, 10)) from stdin;


in multi-row tests , its not correct to use fixed value, should use a column reference

galsalomon66 · 2021-03-26T07:56:05Z

@@ -0,0 +1,4 @@
+select to_string(to_timestamp('2009-09-17T17:56:06.234567Z'), substring(' athmywopgss-nghjkl', 3, 10)) from stdin;
+select to_timestamp(upper('2009-09-17t17:56:06.234567z')) from stdin;


the same as above

galsalomon66 · 2021-03-26T08:03:00Z

+set -x
+set -e
+
+g++ -o queries_generator queries_generator.cpp


its binary utility (cpp sources), it should be a part of project cmake (not need for run.sh)
these utilities need several steps until completion, so its better to add help section to explain that.

galsalomon66 · 2021-03-26T08:06:23Z

@@ -0,0 +1,10 @@
+select lower(lower('  %%AbCdEfGhIjKlMnOpQrStUvWxYz##  ')) from s3object;


some queries containing only constant values which means their results is static (no change between rows)
those kind of queries not need an input object (actually any input object will produce the same result)

galsalomon66 · 2021-03-26T08:10:01Z

+select lower(lower('  %%AbCdEfGhIjKlMnOpQrStUvWxYz##  ')) from s3object;
+select to_timestamp('1999-11-04T20:27:03.479340Z') from s3object;
+select date_add(day, 8, to_timestamp('1982-02-03T00:20:12.42831Z')) from s3object;
+select substring(upper('  %%AbCdEfGhIjKlMnOpQrStUvWxYz##  '), cast((avg(cast(_3 as int)-cast(_2 as int)) - 3) as int), min(cast(_1 as int)-cast(_1 as int)) + 7) from s3object;


👍
this kind of query is unique, since its complex and nested, in such case there is a true need for AWS results to compare.

galsalomon66 · 2021-03-26T08:10:57Z

@@ -0,0 +1,10 @@
+select lower(lower('  %%AbCdEfGhIjKlMnOpQrStUvWxYz##  ')) from s3object;
+select to_timestamp('1999-11-04T20:27:03.479340Z') from s3object;


contain only constant values

galsalomon66 · 2021-03-26T08:11:17Z

@@ -0,0 +1,10 @@
+select lower(lower('  %%AbCdEfGhIjKlMnOpQrStUvWxYz##  ')) from s3object;
+select to_timestamp('1999-11-04T20:27:03.479340Z') from s3object;
+select date_add(day, 8, to_timestamp('1982-02-03T00:20:12.42831Z')) from s3object;


contain only constant values

galsalomon66 · 2021-03-26T08:22:22Z

@@ -0,0 +1,10 @@
+1931-12-23T02:54:25.451925Z


the Z thing cause a failure upon comparing results.
since results are static, and the Z thing has no meaning, it can be removed

galsalomon66 · 2021-06-30T08:53:22Z

+{
+    int num = 1 + (rand() % NUM_COLUMN_INT);
+    aws_expr = "cast(_" + to_string(num) + " as int)";
+    return "int(_" + to_string(num) + ")";


in our current status, there is no difference between aws-expr to ceph-expr.
which is the best thing for comparison tests.
same input queries should produce the same output results.

the "int/float ( ... ) " should remove from the code generator

galsalomon66 · 2021-06-30T08:55:05Z

+            type = rand() % 4;
+            ceph_query = ceph_query + random_query_expr(depth, type,
+                            aws_expr)+ " from stdin;";
+            aws_query = aws_query + aws_expr + " from s3object;";


no need for stdin( it's for specific use-cases, not relevant for this comparison)

Added query generator which generates equivalent queries for aws and ceph. After runing these queries, their results are matched. Signed-off-by: Girjesh Rajoria <girjesh.rajoria@gmail.com>

galsalomon66 · 2021-07-06T16:59:16Z

+    else if (type == TIMESTAMP)  // return type is TIMESTAMP
+    {
+        //switch (option = rand() % 2)
+        //{


why this section is commented?

because if we have to_timestamp function here, it generates simple queries with it. So it is only called depth is zero.

galsalomon66 · 2021-07-06T17:49:19Z

+    fstream query_file, cmd_file;
+    query_file.open("queries.txt", ios::in);
+    cmd_file.open("aws_cmds.sh", ios::out);
+    cmd_file << "#!/bin/sh\nset -x\nset -e\n\n";


another option is to use popen( ... )
https://man7.org/linux/man-pages/man3/popen.3.html
and get rid of the shell

grajoria force-pushed the tests branch from 27aa3cd to a8e6ec6 Compare March 10, 2021 08:44

grajoria force-pushed the tests branch from ad0a46d to 80c0b27 Compare March 24, 2021 11:59

grajoria changed the title ~~Tests for nested queries~~ Tests for nested queries and against aws Mar 24, 2021

grajoria force-pushed the tests branch 2 times, most recently from f3f022a to 3f80232 Compare March 26, 2021 07:47

galsalomon66 reviewed Mar 26, 2021

View reviewed changes

grajoria force-pushed the tests branch from 3f80232 to a0250a6 Compare June 29, 2021 22:46

galsalomon66 reviewed Jun 30, 2021

View reviewed changes

grajoria force-pushed the tests branch from a0250a6 to 9817c8a Compare July 5, 2021 19:28

Test for checking results against aws

03bf4b3

Added query generator which generates equivalent queries for aws and ceph. After runing these queries, their results are matched. Signed-off-by: Girjesh Rajoria <girjesh.rajoria@gmail.com>

grajoria force-pushed the tests branch from 9817c8a to 03bf4b3 Compare July 5, 2021 20:09

galsalomon66 reviewed Jul 6, 2021

View reviewed changes

		@@ -0,0 +1,4 @@
		select to_string(to_timestamp('2009-09-17T17:56:06.234567Z'), substring(' athmywopgss-nghjkl', 3, 10)) from stdin;

		@@ -0,0 +1,4 @@
		select to_string(to_timestamp('2009-09-17T17:56:06.234567Z'), substring(' athmywopgss-nghjkl', 3, 10)) from stdin;
		select to_timestamp(upper('2009-09-17t17:56:06.234567z')) from stdin;

		@@ -0,0 +1,10 @@
		select lower(lower(' %%AbCdEfGhIjKlMnOpQrStUvWxYz## ')) from s3object;

		@@ -0,0 +1,10 @@
		select lower(lower(' %%AbCdEfGhIjKlMnOpQrStUvWxYz## ')) from s3object;
		select to_timestamp('1999-11-04T20:27:03.479340Z') from s3object;

Conversation

grajoria commented Mar 10, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

galsalomon66 Jul 6, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

grajoria commented Mar 10, 2021 •

edited

Loading

galsalomon66 Jul 6, 2021 •

edited

Loading