-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Martin Aspeli wrote:
Tres Seaver wrote:
Please don't deliberately check in failing tests on the trunk. If you need to do this, make a branch, and ask on the mailing list for people to investigate your branch.
Why not? Trunk is (well, was) broken. This makes it clear. The regression actually happened ages ago, but no-one had written a decent test for the original functionality, so the regression was only discovered in application software. This test corrects that omission, and helped a few people co-ordinate addressing the issue.
Why benefit do we get from not making the breakage explicit to everyone?
Two things: - - Most importantly, we have a firm policy that the test should always be "clean" (passing all tests). This means that I don't have to fix the test J. Random Hacker checked in broken before doing work on an unrelated bit of code: I can run the tests before my change, apply and run them afterward, verifying that I didn't break anythin. - - You'll note that "broken" is a matter of opinion here. David actually checked in an update (before my message, but I hadn't read it yet) which indicates that. So, if you think the behavior of the trunk is broken, put your test demonstrating that breakage either on a branch or as a patch in the tracker, and ask folks here to review it. *Don't* bogart the clean build status of your trunk: checking in a breaking test is like pulling the emergency stop lever on the subway. Tres. - -- =================================================================== Tres Seaver +1 540-429-0999 tseaver@palladion.com Palladion Software "Excellence by Design" http://palladion.com -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkuTJAoACgkQ+gerLs4ltQ5/1QCgzjMniDJ29H4FxY0g+jLK2SmL zLMAoLcuonWnoHNOdNYwn0VfBuF5ZiL5 =/oWp -----END PGP SIGNATURE-----