| [m2e-users] Thoughts on M2E |
Hello, I am writing to this list to talk about some concerns that I have about M2E. I've been using Eclipse for years, and both Maven integration plugins before M2E was officially accepted as an Eclipse project. While the quality of the integration was not always good, at least things were going forward to a better integration. So it was quite a satisfaction to see M2E being integrated into Eclipse. Where I work, we saw this as a strong commitment from the M2E people to bring a really great integration between Eclipse and Maven. Recently, I began trying to prepare a new Eclipse distribution for my coworkers. This means downloading Eclipse's latest version, installing the plugins every developpers use and testing that (some of) our projects can be imported and built flawlessly. This has been surprisingly painful to upgrade the M2E plugin. I find astonishly dumb bugs (for example, see [1]) which make M2E look like it's not mature yet. In another situation (see [2]), I was told to use a specific connector but it didn't work out great, to say the least. On thing is sure: the plugin didn't live up to its promises to be a "first class" integration between Eclipse and Maven. But the problem does not only touch features and bugs. As I look though bug reports and the web, I see a seemingly large number of "this is open source, if you want to correct the bug, correct it yourself". Of course, this is open source. Of course, anybody can download the code and correct a bug. But shouldn't the project owners feel somewhat responsible when bugs are found in their software? The feeling I have now is that for M2E, open source doesn't mean "let's share code and build together a better piece of software" but rather "I work on new features and let my users correct the mistakes I've made". Worst of all, I've even found an issue[3] created in the Sonatype GitHub space that asks for compatibility with version 1.8 of Subclipse, which means update the Maven SCM handler for Subclipse. This handler's package is org.sonatype.m2e.*, and allows (among other things, I guess) to materialize projects stored on SVN repositories. In this issue's discussion, I learned several things: - M2E won't support Subversion integration any more, which is a shame given the popularity of this source control software See answer by ifedorenko : "we don't use subversion any more, so this will be WONTFIX unless somebody provides a quality patch. - This doesn't seem incompatible with advertising the SVN integration of the materialize functionnality in the video of M2E's homepage See http://www.eclipse.org/m2e/ - That despite asking the community to provide a quality patch, the patch is not applied once somebody has actually provided the patch. See answer by ifedorenko: "We don't have the time/resources to maintain m2e/subversion integration and won't be able to review and apply the attached patch any time soon. m2e/subclipse integration is provided under EPL, I suggest you fork the project and distribute pre-built binaries yourself (and github makes this really easy to do). " - That despite stopping any work on the handler, and knowing that this handler cannot be installed anymore without errors, nobody took it off the Eclipse Marketplace. When I look at the great things Sonatype has done for the Maven community (we are heavy Maven, Nexus and Hudson users and those are high quality products), I honestly expect more from M2E. So, what now? Should users still trust M2E? Should we expect this integration with the Eclipse organization to bring something useful? Should we switch back to eclipse:eclipse? Or should go one step further and abandon Eclipse as our company's IDE of choice? Best regards, Sébastien [1] https://bugs.eclipse.org/bugs/show_bug.cgi?id=371097 [2] https://bugs.eclipse.org/bugs/show_bug.cgi?id=372847 [3] https://github.com/sonatype/m2eclipse-subclipse/issues/3 |