Precompiled 2024-06-19-dccc63a

Changelog for 1.13 (#27206)

update changelog for 1.13.0-alpha.1

GitOrigin-RevId: f72b348debf0e293d77aad8aeead8cbde064335e

Precompiled 2024-06-18-43fea72

Add the application code for the Fivetran destination connector (#27105)

GitOrigin-RevId: c0776aa6896cf4674e8757850c319e33b72a4363

Precompiled 2024-06-17-28a64af

stop writing to `_module_versions` (#27059)

We're no longer reading from the `_module_versions`  table in prod, so we don't have to write it.

<img src="https://media3.giphy.com/media/c6XTJeqP0mMpjEmd94/giphy.gif"/>

GitOrigin-RevId: 32bdf169a1ee374f8a5b8b071a6ce5029129f34c

Precompiled 2024-06-14-4851928

precompiled-2024-06-14-4851928

Remove unnecessary find_segments_to_compact and CompactionReason::Unk…

Precompiled 2024-06-13-35a4f54

Add text segment cache for opened files (#27002)

This PR adds a `TextSegmentCache` for caching opened text segments. This should reduce the amount of time we spend opening the text indexes, which happens many times over the course of all the RPCs. I changed some of the `*_impl` methods to take in an `Arc<TextSegment>` instead of the segment reader and trackers it needs, so some of the fns have more than they need, but that should be fine since we should be loading everything in the second RPC and getting cache hits in subsequent RPCs for the same search query. We won't open the `TextSegment` on the first RPC (number of segments) because we don't pass the storage keys for all the trackers in, and it's probably not worth optimizing because `number_of_segments` is only needed when we are using single segment format.

GitOrigin-RevId: 8f4166206a546a1e9a3130551dfff9e4559c28d9

Precompiled 2024-06-12-0a47d05

Update tour to use TogetherAI's API instead of OpenAI (#26466)

Due to OpenAI changing their free tier pricing and at least one user commenting that that broke the tour, update Part 3 of the tour to:

- use a simple fetch instead of openai sdk client
- use TOGETHER_API_KEY env var instead of OPENAI_API_KEY
- use one of Together's available models instead of GPT
- update text, filenames & code accordingly  (e.g. 3-gpt becomes 3-ai)

Tested the new convex-tour-chat-3 package in dev, seems to be working as expected

GitOrigin-RevId: cf6bb042f26dd8723ea1b44b7a1683366b0b82a8

Precompiled 2024-06-11-db8d910

precompiled-2024-06-11-db8d910

Refactor out new generic ConfigLoader with separate proto-decoding im…

Precompiled 2024-06-10-dc316ab

Strip unused dependencies based on cargo machete (just machete) (#26867)

Remove bunch of unused dep arrows in the dep graph.

GitOrigin-RevId: e6d9c293231dadcbbba241b40519614f3058103c

Precompiled 2024-06-07-00f5ec1

run SchemaWorker in all namespaces (#26817)

run schema worker in every namespace that has a `_schemas` table

GitOrigin-RevId: e1a9776a68c7c9bed8300f682c62f41c30bc28f5

Precompiled 2024-06-04-17f0813

Add bigbrain support for searchlight clusters (#26562)

GitOrigin-RevId: 692cb9394db34b42015f5f784868086ebe7c651b