tag:blogger.com,1999:blog-20038672.post2565915794063071518..comments2024-03-28T00:58:29.187-04:00Comments on Robert Haas: PostgreSQL Regression Test CoverageRobert Haashttp://www.blogger.com/profile/08393677427643988650noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-20038672.post-24823418740862315532016-05-27T12:41:42.664-04:002016-05-27T12:41:42.664-04:00You could take a look at "mutation testing&qu...You could take a look at "mutation testing" [https://en.wikipedia.org/wiki/Mutation_testing] to answer the question if a certain test/test suite has real value. It involves the following steps:<br />- Modify the code under test<br />- Run the test suite (or just parts of it)<br />- Look if at least one test failed<br />If no test failed, you have found code that has not been tested at all. So this goes one step further than "code coverage" which just tells you that a test has executed certain code.<br />Additionally, the same procedure can be used to compare two tests. If the fail for the same mutations, the are equivalent. You could then reduce the test suite by removing test cases that do not add any value.zimmskihttps://www.blogger.com/profile/10145704632149180728noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-74272382908734939572016-05-26T20:15:29.581-04:002016-05-26T20:15:29.581-04:00Some of the drop between 9.4 and 9.5 is most likel...Some of the drop between 9.4 and 9.5 is most likely because check-world in 9.5 will only create one temporary installation instead of one for each test subdirectory.Anonymoushttps://www.blogger.com/profile/02849480732923051923noreply@blogger.com