meltano hub, could fill the gap of answering “where is the most active fork of this tap”, but that doesn’t solve discovery/install problems like google search and pypi references

That community intervention is what we’re aiming to solve with the Hub! I think we’d have a couple of approaches with this one. First, we could make the upstream for those PRs the default variant until it’s actually merged. That person may not want to take full ownership of it though so some other options we’ve toyed around with are having a generic “meltano community” namespace on github that community members can become maintainers on as needed. If someone says, I don’t want this under my personal namespace but I’m happy to support reviewing MRs, we could support that workflow and point the hub to the shared namespace.

yeah, i think that’s the main thing. in this case, the maintainer just needs to do a review and merge it in (probably). The more we can automate testing and conventions with the SDK, the lighter (and more generalizable) that becomes as well.

Right - the big challenge I see, and i don’t have an answer for this, is how to test against an API that we (the broader community) don’t have access to.

right, that might be the biggest challenge. a lot of the taps i see developed i have never even heard of. so specific responsibilities would be; • access to the source system and willingness to test new PRs ( if not automated) • willingness to review and merge PRs • willingness to tag new releases / deploy to pypi (or do we just assume people are forking at certain commits?)

yeah, makes sense. I’m not sure how the PyPI situation would work. Since Meltano could manage the aliasing to different variants. Basically adding

hub/tap-whatever

would always be the latest variant

@taylor and @stephen_bailey - While I have seen good success with the

dataops-tk

community/decentralized approach so far, I want to make sure we’re making a considered decision on this as we experiment to open up to more community developments. Meltano of course has limited bandwidth and we want to make sure we’re making the right long-term support decisions for our future growth and also for the longterm needs of the community. I’ve logged a very detailed writeup of what I see as tradeoffs and consideration in the various support models we have available today. Thoughts and suggestions welcome. 👍 Add option for community management of singer forks (#73) · Issues · meltano / Meta · GitLab

@aaronsteers Thanks for filing this AJ! @stephen_bailey Would you mind weighing in on the issue with your perspective? It sounds like you’ve spent some of your own brain cycles on this issue already 🙂

Will do! I actually drafted a response a while back that turned into an essay, which I deleted 😂

Haha, I know how it goes