Ansible.netcommon broken with Ansible Core 2.19.0

The concerned PRs are merged, and we are planning to push a release by Thursday (Aug 14)
While we are planning the release, I would like to thank @mattclay @nitzmahone @Ruchi_Pakhle @gundalow @komal for all the help and support in getting things fixed and running.
Regards

5 Likes

ansible.netcommon v8.1.0 is released :tada:
The underlying network platform collections would have a iteration of major releases to update the known dependency (planned next week)

Regards.

2 Likes

@KB-perByte I’m not quite sure what this means. Does ansible.netcommon v8.1.0 fix the issues with ansible-core 2.19.0, or will the collection be compatible with ansible-core 2.19.1+ only?

I’m sorry if you’ve already answered this somewhere else. I wasn’t able to find this :frowning:

BTW the meta/runtime.yml states requires_ansible: ">=2.16.0". Out of curiosity: If the collection will be compatible with 2.19.1 but not 2.19.0, would there be a way to define this? Something like requires_ansible: ">=2.16.0;!2.19.0" or similar?

@KB-perByte why are you bumping major versions for the network collections? I checked prep release 11.0.0 by KB-perByte Ā· Pull Request #1225 Ā· ansible-collections/cisco.ios Ā· GitHub as an example, and I see zero reason to make a major version bump. The only changes seem to be a bugfix and a dependency bump, which can both be done in a minor release. Same for prep release 12.0.0 by KB-perByte Ā· Pull Request #559 Ā· ansible-collections/cisco.iosxr Ā· GitHub, there’s no reason to make this a new major release.

Please reconsider and make minor releases!

Is is really acceptable to do a dependency bump to anothor major version in a new minor version? I’m always unsure about this and try to avoid it.

Anyway, I’m not even sure if this dependency bump is really necessary. I’ve commented in the PRs.

I think it’s OK (that doesn’t say anything about whether it’s a good idea), since the package manager (ansible-galaxy) has a resolver that takes care of dependencies.

Whether it’s a good idea: I agree with you that it’s not necessary, and I personally wouldn’t do it either. Bumping the minimum required dependency version without this being really required only makes it harder for many users who cannot update to that dependency for whatever reason.

Also ansible-galaxy is still missing some important features, like making use of requires_ansible (that was originally planned for ansible-core 2.19, but got moved to later), which can make this bump more problematic for users if ansible.netcommon drops support for ansible-core versions with that bump.