Wouldn't something simplistic like squeak PackageDependencyTest help? Nicolas
Le 26 février 2012 09:58, Stéphane Ducasse <[email protected]> a écrit : > >>> Of course it is always much better if the author of the patch can see >>> that it caused some problem than when some other person must later >>> investigate what patch of an update has evil side effect. >>> >>> I hardly need to tell you that it is very demotivating to wait what >>> next will break something again (as this week one single update fixed >>> one issue and brought the next one). Instead of improving of the >>> system we must spend a lot of energy on maintenance. >>> >>> We definitely must improve the QA process. >> >> To support my lemmenting I must say that the state of non-failing >> kernel jobs lasted only 10 hours. :-) > > Excellent! :) > Seriously it means that we should work on rules checking so that we identify > the violations. > People used tests for that now as soon as we have small lint we should be > able to express architecture rules. > > Stef > > >
