Hello all!
I would argue that in addition to #1534, #1524 should definitely be part of 3.0. The ability to control who has access to certain documents is something most users would expect in 2019 at this point. Right now there's really no way to do this without putting a convoluted proxy to act as per-document access control in front of your app. >From a user perspective a 2.X to 3.X upgrade should result in a pretty >substantial upgrade in functionality and not necessarily code improvement >(Erlang 22 support, and Elixir test suite for example). There are other things >I'd recommend from the backlog but as to note dilute my personal ask, I'd say >per document access control is a must as pretty much everyone I know who's >using CouchDB seriously in a situation that's not neatly set-up to fit a >database-per-user will end up inevitably implementing this themselves. Tabeth P.S. As an aside, has there been thoughts on scheduling upgrades on a time basis as opposed to feature basis, ala EmberJS or Ubuntu? What this would mean in practice is that at some cadence, e.g. 4 times a year everything done at that point would be wrapped up in a new version. I think this would help the community in that there's a steady march towards functionality being available for production use. This would also tighten the feedback loop between releases and feedback. ________________________________ From: Joan Touzet <woh...@apache.org> Sent: Wednesday, August 7, 2019 11:52:01 AM To: CouchDB Developers <dev@couchdb.apache.org> Subject: Getting started on CouchDB 3.0, and an introduction Hi everyone, Now that we have a path forward for FoundationDB, we also need to get moving on our best-and-greatest CouchDB-as-is release, namely the 3.x branch. If we were to cut CouchDB 3.0 from master today, we'd already have a lot of great new things since 2.3.1: * Partitioned DBs * Open PRs: https://github.com/apache/couchdb-documentation/pull/385 * Cloudant-donated industrial-strength IOQ replacement * Cloudant-donated compaction daemon replacement ("smoosh") * Cloudant-donated view warmer ("ken") * Ready for Cloudant Search, if installed separately (no recompile) * Native shard splitting functionality * Improved test suite (JavaScript -> Elixir) * Erlang 22 support * Fix for rare fsync error encountered by Postgres in 2018 But we planned for a few more things for 3.0, as you can see in the "Proposed 3.x" column here: https://github.com/apache/couchdb/projects/1 It would be good for us to decide which of these are making it into 3.0 itself. At a minimum, we need #1534. I am actively working on #1523, and I know Jan is actively working on #1524. There's also the backlog column - we should determine if any of those make sense for 3.0 as well. I see a few I'd love to have in there personally, but don't have the time to commit to them just now. --- Also, allow me to introduce Deni Burroughs, who has been lurking on the list for a while now. Deni is the dbcore Cloudant Engineering Manager, and she'd like to help coordinate getting 3.0 out the door. Her area of expertise is in non-technical project management, which is great, because we could use more help in keeping our cats corralled! :) She and I had a chat a few weeks ago, and now seemed like the best time to introduce her to the list. Please make her feel welcome! -Joan
