Greetings, On Wed, Feb 15, 2012 at 8:15 AM, Stephen Connolly <[email protected]> wrote: > IOW, the command shown on the console log will not behave the same if > the user executes it themselves from the same directory as the same > user. > > EVIL
I think there is some misunderstanding about what the point of the Maven plugin is. It isn't simply to invoke Maven. It is to bind Jenkins with Maven project in a reasonable way. Therefore, I think your claim of EVIL is spurious. If a Jenkins job requests checkstyle reporting, then a Maven job type which ensures checkstyle is run is the correct behavior. It will not be a complete match up to what command lines were invoked for the build itself - it would be WRONG for Jenkins not to have different behavior. > I have other issues with the Maven Project type. but "Modifying the > effective pom" is the most evil one of all I'm not sure if these two are roughly the same. Again, your attack against Maven job type is highly suspicious to me. Most times, when I see someone having a problem with a Jenkins Maven build it is (like you observed) a Maven problem. Of the remaining ones, I also ask them to try it as a Freestyle job type, but I go further than you (I think) in that I request they pass the effective command line flags which would have been invoked by all the configuration options which they have enabled (e.g. Checkstyle) for their Maven job type. The majority of these times are still broken! That is, the Freestyle job type is not a magic solution, it is almost always a misunderstanding by the user of what is actually going on during the build. I think the Jenkins Maven support comes under a lot of attacks which it doesn't deserve. -Jesse -- There are 10 types of people in this world, those that can read binary and those that can not.
