Time for GeoParquet 1.0?

[cholmes]

I'd like to propose we release the first GeoParquet 1.0 'release candidate' in the next couple weeks (after getting in any open PRs), and then if there are no bugs raised against it within a month then we call fully release 1.0.0.

We have not quite achieved all the targets we discussed in the path to 1.0.0, but I believe we're in a bit of a 'chicken and egg' with some providers / implementors waiting for the 1.0 before they make a push to do it. I'll detail how close we are below on the various aspects we said, but there has been progress on a number of fronts. The other aspect in play now is that GeoArrow and its columnar geometry type is also waiting for GeoParquet 1.0, which we plan to bring in with GeoParquet 1.1. That is where GeoParquet likely really starts to shine and get increased adoption, so we should aim to get the interoperable core metadata out soon, and then work for more adoption after that.

In terms of progress, where we've fully met our goals:

• We're doing great on implementations, with roughly 15 between 'tools' and 'libraries' listed on https://geoparquet.org
• We've got a core website up at https://geoparquet.org that publishes the spec, list implementations and gives a basic overview.
• There's a complete validation engine as part of gpq that tests the data and metadata. And making it an online validation tool is in progress.
• Progress has been made with OGC, with a 'Standards Working Group' in formation, which will allow us to be an official OGC standard faster than doing a community standard.

We've made progress in areas that are a bit short of our goals:

• We aimed for 5 'data providers', and right now there's really only 2 - microsoft and source cooperative. But there are discussions with a few more and we can use rc.1 as a push to get a few more to be included in the announcement.
• We aimed for 20 different 'production datasets', and we've got arguably 5 or so (unless you count each Planetary Computer STAC catalog as geoparquet as it's own, which seems a bit too easy). But I think we'll see more (I personally worked on 2 for source, and will try to put up a couple more useful but easy ones).
• Test data. We do have a PR up at Add test data covering various options in the spec #134 that proposes some. It would be great to get this in, but it does not feel absolutely essential to get to 1.0

In conclusion, for me the truly core things were:

1. Get a website up
2. Have a validation engine that would check not just metadata but data, and GPQ does that, and I think there is also a GDAL/OGR based one that can do it too.
3. Get a few more implementations and more 'thumbs up' from people looking at it. For the later we've had a number of positive conversations with most all of the major providers and projects in the space. We had a flurry of discussion around making metadata optional and WKT vs WKB which helped question from some more angles and lead to compatibility recommendations PR.

So my proposal is to get in the compatibility recommendations PR and then cut RC.1.

Thoughts from others?