| |||||||||||||
Bugfixes introducing bugsBugfixes introducing bugsPosted Dec 4, 2008 10:25 UTC (Thu) by NAR (subscriber, #1313)Parent article: MySQL 5.1 and development models This in combined with a failing review process have introduced of a lot new bugs while trying to fix old bugs. This is something that an automated test suite should prevent...
(Log in to post comments)
Bugfixes introducing bugs Posted Dec 4, 2008 19:19 UTC (Thu) by droundy (subscriber, #4559) [Link]
I would say only that it's something that an automated test suite can help with. Writing a complete test suite for a relational database most likely takes far more work than writing a relational database. The problem is that new bugs usually show up in odd corners which are unlikely to be tested by an automated test suite.
Bugfixes introducing bugs Posted Dec 6, 2008 17:33 UTC (Sat) by kleptog (subscriber, #1183) [Link]
While that's true, PostgreSQL's extensive regression test suite does turn up an amazing variety of bugs. Together with the buildfarm which runs the tests over a large range of servers and operating systems. Over time it has revealed a few fairly subtle bugs.
I suppose they had the advantage that the test suite has been there since the beginning and the enforcing the rule that every feature must come with a regression test keeps it up to date.
I think what really clinches it though is the number of people willing to test new versions prior to release. Every release there's generally a must have feature for people running expensive servers with large datasets and who throw insane queries at it.
(My favourite proposed regression test was using WITH RECURSIVE to generate the Mandelbrot set in SQL. Amazing piece of work, though quite hard on slow machines.)
|
|||||||||||||