Re: Modulesync Config 4.0.0

David Schmitt

Given that puppet5 is EOL RSN I wouldn't hold my breath for any new platforms being supported in that stream. Similarly with puppet7 being released, I expect little investment in adding new platforms to puppet6 going forward.

Cheers, David

On Sun, 29 Nov 2020 at 21:11, Tim Meusel <tim@...> wrote:
Hi people,

thanks for all the awesome work Ewoud!

Tiny update here: you might see failing tests on Ubuntu 20.04 with
Puppet 5. That's because Puppet does not provide packages for this
distro. The same applies for Fedora 33 on Puppet 5/6 and Fedora 32 on
Puppet 5, but we rarely have those in our metadata.json.

We already merged a few PRs where *only* acceptance tests fail for the
above mentioned platforms. IMO it's fine to continue with this until
Puppet provides packages / we figure out how do exclude those OS/Puppet
combinations from our test matrix.

cheers, Tim

On 26.11.20 13:37, Ewoud Kohl van Wijngaarden wrote:
> Hello everyone,
> You may have seen a wave of activity yesterday. Bastelfreak and I
> released our msync config 4.0.0.
> The major change was from Travis CI to Github Actions. In doing so, the
> test matrix has become dynamic. This means that metadata.json drives it.
> If you list Puppet >= 5.5.8 < 7.0.0 it will run the unit tests on Puppet
> 5 & 6. To add Puppet 7 support, just update metadata.json and it'll run
> the tests. Similar, Puppet 5 can be dropped in the same way. We found
> some modules which didn't declare any support or still only Puppet 4.
> Most of these have been cleaned up.
> Similarly, if there are acceptance tests it will run them. That means
> that if you list CentOS 6 in metadata.json and there are accepance
> tests, it will run them. We found a few broken modules and they have
> dropped CentOS 6. There are also a bunch of modules where the acceptance
> tests all fail because previously they didn't run. There are still a
> bunch of them open now.
> Now there are 52 open modulesync PRs.
> For each one I left a message why they failed but I don't have time to
> fix all issues. This is where you come in.
> For a few Github Actions didn't trigger for some reason. I'll see if I
> can work on that.
> Then there are modules where there are problems that need code changes.
> Either in the modules or in the tests. These have the tests-fail label.
> I'd like to ask everyone to look at the modules they care about and fix
> them. I'll be happy to advise but often I don't even know the software.
> If modules are really broken for a longer time and nobody steps up I'd
> suggest to mothball them. This means removing it from the modulesync
> config so they're not holding us back. They should also have an open
> issue which describes the state of the module.
> Regards,
> Ewoud Kohl van Wijngaarden

Join to automatically receive all group messages.