Hi all,
This is about moving support for network devices out of core and encouraging packaging this content in Galaxy instead.
This won’t include load balancers - In some cases, for things like load balancers that are widely utilized in Ansible land, I think it makes sense to continue to have support for them in core. These things are widely used enough that will receive frequent patching, and usually virtual machine implementations are available to more or less easily set up test environments.
However, for some hardware, such as networking gear, I’d really like to see these packaged in galaxy instead, leaving support up to the vendor.
Many Ansible users are cloud based and won’t be needing these modules, and usually support is coming from one place, rather than a common shared project that we are moderating. This also allows others to moderate those improvements and progress at their own rates, as well as to not have to adhere to our particular conventions (should they wish), including that certain modules need to be written in particular languages.
As such there can be a Galaxy role like “arista_modules”. In this role, if the role has a subdirectory called “library/” it can package all the modules it needs within them, and the content on GitHub can be updated on GitHub at any time without relating to a core release.
We’ll continue to take vendor modules in core for all sorts of proprietary services still - including monitoring services and notification modules - if they are going to be easy to keep up and test - though it seems networking equipement should not be one of those things, since it’s much more involved and not always in the main use case path for what Ansible is or does.
Simply put, vendors can do a better job maintaining their equipment, and we don’t want to sit as gatekeepers in front of that, as it’s not a core use case for Ansible.
That being said, we may choose to suggest some other modules also live in Galaxy in the future, but we have no plans to move other things out of core in the midterm - though some new module contributions of other kinds, should we may not choose to include them, we may suggest be held in Galaxy instead.
Folks should still keep submitting to core first - as we want to consider having as strong of a core as possible, just that we do wish to keep breath under control a bit to keep the overall list manageable.
Hope this makes sense
–Michael