Will, you make a fair point and we should not be removing plugins just because they fail to build but that's NOT what I've said.
Let me break down my arguments: - The only obligation project has is towards any CloudStack users who may be using these plugins, but given the state of the plugin it's highly unlikely that they are in production use. The purpose of this thread is to investigate and ask if there are any such users, so far I'm not hearing anything from any of those users. - If the vendors who had initially contributed the plugins are not maintaining them or are not responsive, the project should not be obligated towards maintaining a broken component that does not even build, and project should in that case work towards a plan to deprecate such plugins over time. - The first thing I'm proposing here is to comment those plugins in 'plugins/pom.xml' to exclude them in the default build process. The next steps could be to discuss deprecating and removing them from the codebase over time, this is open for discussion and should be discussed separately. - The specific plugin (contrail) also fails to build against JDK8 that adds a roadblock to our plan to migrate to JDK8 in future. - Background: I checked with few people including original authors/contributors, the story I'm told is that several of the network plugins were created as a proof-of-concept or go-to-market tools, and did not take off or got attention from their vendors as they failed to achieve specific business goals. Given CloudStack has been user-driven (than vendor-driven) it is fair to conclude that several of the plugins are not maintained most-likely because nobody is using them. Regards. ________________________________ From: williamstev...@gmail.com <williamstev...@gmail.com> on behalf of Will Stevens <wstev...@cloudops.com> Sent: 27 October 2016 22:50:19 To: dev@cloudstack.apache.org Cc: Rohit Yadav; us...@cloudstack.apache.org Subject: Re: Disable open inactive plugins: Contrail plugin Just because recent builds are failing does not really mean that no one is using it. In my experience working with different companies who have ACS in production, a lot of them are using much older versions of ACS (4.4 for example). Only a subset of companies keep their ACS install "close" to master and they are likely 2 or 3 versions behind master as well. I would suggest we wait a bit to see if anyone from the users@ list pops up. I think we can probably disable Midonet. I think Contrail is more likely to have active users on previous versions. I would be in favor. I think that nobody uses them since all recent builds are failing, right? Your proposal seems good to me. Wido > > Regards. > > rohit.ya...@shapeblue.com<mailto:rohit.ya...@shapeblue.com> > www.shapeblue.com<http://www.shapeblue.com> > 53 Chandos Place, Covent Garden, London WC2N 4HSUK > @shapeblue > > > rohit.ya...@shapeblue.com www.shapeblue.com 53 Chandos Place, Covent Garden, London WC2N 4HSUK @shapeblue
