mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2026-10-28 16:18:23 +08:00
Konstantin Bogdanov
023c1c44e5
The performance CI runs every test with `--max-queries 10`, so a test that expands to more queries than that runs only a random sample on each CI run. `math.xml` expanded to 60 queries, so a regression in any one function had roughly a one-in-six chance of being measured per run. Split it into 8 files of 6 to 9 queries each, grouped by function family and argument domain, so every query runs on every CI run. The set of queries is unchanged.
23 lines
660 B
XML
23 lines
660 B
XML
<test>
|
|
<substitutions>
|
|
<substitution>
|
|
<name>func</name>
|
|
<values>
|
|
<value>log</value>
|
|
<value>log2</value>
|
|
<value>log10</value>
|
|
</values>
|
|
</substitution>
|
|
<substitution>
|
|
<name>arg_positive</name>
|
|
<values>
|
|
<value>toFloat64(number + 1)</value>
|
|
<value>toFloat32(number + 1)</value>
|
|
<value>number + 1</value>
|
|
</values>
|
|
</substitution>
|
|
</substitutions>
|
|
|
|
<query>SELECT count() FROM numbers(100000000) WHERE NOT ignore({func}({arg_positive}))</query>
|
|
</test>
|