[SOLR-17189] DockMakerTest.testRealisticUnicode fails from whitespace assumption

• Type: Test
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 9.6
• Component/s: benchmarks
• Assignee: David Smiley
• Reporter: David Smiley
• Created: 01/Mar/24 04:19
• Updated: 28/Apr/24 19:29
• Resolved: 10/Apr/24 12:52

Description

DockMakerTest.testRealisticUnicode fails 1-2% of the time – link.

java.lang.AssertionError: expected:<6> but was:<7>
at __randomizedtesting.SeedInfo.seed([C5136F274AFF3ADD:95FFEBF499446D74]:0)
•••
at org.apache.solr.bench.DockMakerTest.testRealisticUnicode(DockMakerTest.java:189)

It seems clear it's because it assumes that the "realistic unicode" chars won't match the regexp: \s. A single space char is used to join the words but maybe this or other whitespace chars are in those unicode codepoint blocks.

Additionally, it's frustrating that this particular benchmark framework doesn't honor tests.seed in its generation of random data and thus it's hard to reproduce the failure. That ought to be fixed as well.

Issue Links

• links to: GitHub Pull Request #2327