Repository navigation
fix: add docs snippets; fix open api issues - #1913
Merged
Merged
Conversation
szuperaz
commented
Oct 15, 2026
| custom: {}, | ||
| // The API reads the AI indicator fields from `custom` and puts them on the event it | ||
| // delivers. They aren't app custom data, hence the assertion. | ||
| custom: { |
Contributor
Author
There was a problem hiding this comment.
This one threw me off at first, but I verified that, for ai_indicator.update events, the message_id and ai_state fields need to be wrapped in custom, but then when the backend broadcasts the event, it will read these fields from custom and add them to event.message_id and event.ai_state.
szuperaz
marked this pull request as ready for review
October 15, 2026 20:31
szuperaz
requested review from
MartinCupela,
isekovanic and
oliverlaz
as code owners
October 15, 2026 20:31
oliverlaz
approved these changes
Oct 16, 2026
github-actions Bot
pushed a commit
that referenced
this pull request
Oct 20, 2026
## [10.0.0-rc.23](v10.0.0-rc.22...v10.0.0-rc.23) (2026-10-20) ### Bug Fixes * add docs snippets; fix open api issues ([#1913](#1913)) ([53505e7](53505e7)) * address code issues found while reviewing v10 source comments ([#1922](#1922)) ([2fc319e](2fc319e)) * v10 in depth review hotfixes ([#1920](#1920)) ([98a298d](98a298d))
|
🎉 This PR is included in version 10.0.0-rc.23 🎉 The release is available on: Your semantic-release bot 📦🚀 |
1 of 2 tasks
isekovanic
added a commit
that referenced
this pull request
Oct 21, 2026
## CLA - [ ] I have signed the [Stream CLA](https://docs.google.com/forms/d/e/1FAIpQLScFKsKkAJI7mhCr7K9rEIOpqIDThrWxuvxnwUq2XkHyG154vQ/viewform) (required). - [x] Code changes are tested ## Description of the changes, What, Why and How? Since #1913, `getThreadAndHydrate` returns the thread registered in `client.threads` instead of the copy it builds, and skips hydrating it when the thread is open. `thread.reload()` went through `getThreadAndHydrate`, so an open thread got itself back and `hydrateState(this)` returned early. Connection recovery reloads every open thread, and none of them caught up: replies sent while the user was offline never appeared, even after closing and reopening the thread. ### Reload `thread.reload()` calls `client.getThread` and hydrates itself with the answer, merged into the loaded window as before. It no longer depends on the thread being registered. ### Hydrating from a response `thread.hydrateState` takes the server response (`ThreadStateResponse`) instead of another `Thread`, like `client.hydrateChannels` and `client.reminders.hydrateState` take data. The mapping from response to state is shared with the constructor (`formatThreadState`). `skipReplies` updates the state and leaves the replies and `isStateStale` alone. ### Queried threads `ThreadManager.resolveQueried` takes the response instead of a copy. It returns a new `Thread` when none is registered, otherwise the registered one with its state updated: - open: replies left to its own `reload()` - stale and not open (every closed thread after a reconnect): replies replaced by the answer - current and not open: replies left alone, events kept them current Queries no longer build a copy for threads that are already registered. Each copy used to keep its replies held in the shared message store, so every thread list reload pinned the listed threads' newest replies for good. `ThreadPaginator.toStoredInstance` is removed, since `queryThreadsAndHydrate` already returns the stored instance. ### Behaviour changes - A registered thread that is open, or not stale, now gets its state (title, read state, reply count, participants, parent message) from the query. Before, it was returned untouched. - Thread responses no longer refresh the channel's `custom` data and members for threads that are already registered. That only happened as a side effect of building the discarded copy. - A second `ThreadManager`, or a `ThreadPaginator` with a private store, no longer dedupes queried threads against its own store. ### Tests The reload and thread list tests stubbed `getThreadAndHydrate` / `queryThreadsAndHydrate` with ready-made `Thread` copies, which is why none of them caught the regression. They now stub the raw `getThread` / `queryThreads`, so the registry lookup runs. New tests cover reloading a registered open thread and the three `resolveQueried` cases. ## Changelog - Fixed `thread.reload()` not applying the fetched replies to an open thread, so open threads catch up after reconnecting - Fixed thread queries leaving the replies of already registered threads held in the message store - `thread.hydrateState` now takes a thread response (`ThreadStateResponse`) instead of another `Thread` (breaking) - Thread queries now update the state of registered threads in place, including open ones
github-actions Bot
pushed a commit
that referenced
this pull request
Oct 22, 2026
## [10.0.0-rc.24](v10.0.0-rc.23...v10.0.0-rc.24) (2026-10-22) ### ⚠ BREAKING CHANGES * remove messageSetToAddToIfDoesNotExist and let the message paginator place every page (#1926) * open api hot fixes (#1927) ### Bug Fixes * apply thread reloads and queries to the registered thread ([#1929](#1929)) ([acb33c4](acb33c4)), closes [#1913](#1913) * open api hot fixes ([#1927](#1927)) ([4fb4b52](4fb4b52)) ### Features * remove messageSetToAddToIfDoesNotExist and let the message paginator place every page ([#1926](#1926)) ([2b7960d](2b7960d))
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
https://linear.app/stream/issue/REACT-1069/migrate-client-side-chat-qa-tests-to-v10-of-stream-chat-js
Migrates the docs-snippets test project from v9 -> v10. It contains all docs and platfrom stream-chat-js docs snippets we have. Relevant docs branch: https://github.com/GetStream/getstream.io/pull/683 (will be kept in sync with
getstream.io/mainuntil we're ready to release stream-chat-js).Most changes are to
docs-snippetproject, andsrc/gen, however, some client issues were found by the tests and those are fixed as well, so PR title isfix.Fixed client issues
Most changes are type fixes, 1-2 behavioral; here are the most important ones:
queryThreadsAndHydrate/getThreadAndHydratenow return the already-registeredThreadthrough the newThreadManager.resolveQueried. Before, they built a second instance thatactivate()and incoming events didn't reach. A registered thread that is out of date and not open gets refreshed from the query result.channel.sendEventmethod expected the response Event type, instead of the request. So it expected fields likecreated_at,custom. Fixed it, this method now expects the event request typeclient.devToken()is back (I think this was mistakenly removed with other server-side stuff)TODOs
Apart from the client changes in this PR, we have 3 small, pending todos to generator/open api spec. All tracked in the docs-todo file. The run also caught a bunch of smaller API behavior issues; since these issues are present on api v1 and v2 as well, they're not blocking us. I'll verify the findings and pass them to the backend team once I have the time.