🚀 Announcement: RxDB v17 - Next Major Release

[pubkey]

We are excited to announce the upcoming RxDB v17, a new major version focused on improved clarity, better tooling, and a more consistent product offering. While v17 is a relatively small major bump, it introduces important refinements that help prepare RxDB for future features and long-term stability.

This release does not include any large-scale architectural rewrites. Instead, v17 focuses on polish, naming consistency, and improved visibility into database internals.

All planned changes for the v17 release are documented here:

👉 https://github.com/pubkey/rxdb/blob/master/orga/before-next-major.md

I'd Love Your Ideas

As I finalize v17, I'd love to hear your suggestions.
Is there anything I should improve, simplify, or reconsider before the release?

[pubkey]

Just released the beta version, please test: https://www.npmjs.com/package/rxdb?activeTab=versions

[tsekiguchi]

I'd like to raise a concern over limiting the size of the string to 500 char. I'm using SQLite and currently I'm indexing on file paths, which can be totally variable based on a user's preference for the names of folder, and how many folders and subfolders they use before actually getting to the path of the file. I originally had it set to 2048 max, and even then I ran into an edge case issue. If I'm no longer going to be able to index on a path this long, then what's the recourse action? Is there a happy medium where we aren't at 10,000, but find a reasonable max that won't break other storages? or is there a way to denote which storages will break with indexing over 500 characters?

[pubkey]

@tsekiguchi at the moment the limit is 1000. Should 2048 be ok?

[tsekiguchi]

@pubkey I think 2048 should be okay for most, but maybe we can be more deterministic. Which storage has problems when using a maxLength of 10,000? Is there a specific maxLength number at which storages start running into problems?

[pubkey]

@tsekiguchi Set to 2048 here: d3ac3a0

There is no storage that fails on a specific lenght. But some storages like memory and OPFS get slower and slower the longer the indexed key is. This check is mostly added because people used chatGPT to generate their schema with a default lenght of something really big like 1000000 which then fully blocks the javascript process on writes. I do not see any real reason to index such long strings. All edge cases can be solved by indexing a hash or a part of the string instead.