AK's weblog

Dear Sam,
Although what you describe comes quite close to CI, it isn't. Unless you compile the code and run the test suite in a clean and neutral build environment, you can never be sure that the changes you're about to commit really do integrate properly. Even if you do a svn/cvs update (or the equivalent command of your favorite SCM) before you compile, run the tests and commit, you'll never know whether it compiles in other build environments. Perhaps you've forgotten to add a file to be committed, perhaps you've relied on certain specifics of your local system when tweaking something on the Makefiles. People make these mistakes all the time, and that's why we shouldn't rely on people running test suites locally.

That's what CI system like CruiseControl or Hudson are there for: they rebuild the software and run the test suite in a clean environment (ideally) whenever a commit is done, in continuous iterations, automatically, without any manual intervention by the programmer. With such simple little helpers combined with some notification, such integration mistakes or any other regressions that can be caught with the existing test suite wouldn't go unnoticed. Hey, you could even send notification emails about build and/or test failures (including their recoveries) to a public mailing list, which would dramatically increase the visibility of such automatized efforts.

If you already do so, I apologize for my possibly rude tone, but your article suggests that you don't.