Split out google connectors into its own repo

[mdedetrich]

So due to the fact that I have started working on google cloud storage and there will be feature/behaviour changes which asks for a 1.2.x release I was wondering whether it makes sense to start our journey on splitting out various connectors from this pekko-connectors mono repo and that we can start with google cloud storage.

Specifically with google cloud storage, I was thinking that we should actually have 2 repos. One which stores google-common (for those who don't know, google common has all of the shared code for all of the other google related connectors, primarily auth related) and then we would have another repo for each other google component, i.e. google-cloud-storage, google-bigquery etc etc.

Just to be clear, this would be primarily moving code around. There won't be any breaking binary changes (which can be confirmed with MiMa) and all artifact names etc etc would remain the same

wdyt?

[raboof]

I'm not opposed to this change. One consideration is that this means these components will have their own release/vote cycle, right? That might be challenging as it's already sometimes difficult to find people to vote on releases. On the other hand, reducing the scope of the release might make it easier.

(related: I plan to work on moving more of the release preparation process to CI later this year, that might make things slightly easier as well)

[mdedetrich]

I'm not opposed to this change. One consideration is that this means these components will have their own release/vote cycle, right?

Yes that is the main point actually, I want to avoid doing a 1.2.x release for pekko connectors when in reality only 1-2 connectors have major changes

That might be challenging as it's already sometimes difficult to find people to vote on releases.

I think this would actually make things easier, right now I think people are more or less blindly voting for releases especially in the case of pekko-connectors where its not entirely obvious which connectors have zero changes and which ones have actual changes.

On the other hand, reducing the scope of the release might make it easier.

This is how I see it.