mirror of
https://github.com/stack-auth/stack.git
synced 2026-07-20 21:29:36 +08:00
32daff06f5
### Context
We were seeing the txn table and the customers tab under product page
OOM.
It turns out this was because the team icon when loaded would fetch all
of the teams.
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Fixes OOM and renderer crashes in the transactions and product customers
tables by stopping per-row refetch storms and avoiding list‑all‑teams
calls. Infinite scroll is now robust, and shared avatar skeletons keep
table rendering smooth.
- **Bug Fixes**
- Data grid (`@hexclave/dashboard-ui-components` `use-data-source`):
queue `loadMore` while a fetch is in flight and replay only after a
successful settle; discard queued requests when pagination mode leaves
infinite; commit the cursor only after a result is delivered; skip
redundant refetches when inputs match a completed fetch; abort in‑flight
requests on unmount.
- Transactions table and product customers tab: wrap `User*`/`Team*`
avatar cells in `Suspense` with a shared `AvatarCellSkeleton` to prevent
per‑row fetch storms and suspend thrash.
- `serverApp.getTeam`/`useTeam` (in `@hexclave/shared` consumers): fetch
a single team by id via a cache; return null for invalid ids; keep
user‑scoped variants membership‑scoped; stabilize hook order. This
removes the “fetch all teams per row” behavior that triggered OOMs.
- Prefetching: cap `/projects/*/teams` to `useTeams({ limit: 1 })` and
remove the heavy owner‑team users prefetch.
- **Refactors**
- Add unit tests for infinite pagination in `use-data-source` (deferred
`loadMore`, aborted resets, error handling, and cursor continuity).
<sup>Written for commit
|
||
|---|---|---|
| .. | ||
| src | ||
| package.json | ||
| README.md | ||
Hexclave SDK Specification
This folder contains the specification for Hexclave's SDKs.
When writing this specification, try to write imperative pseudocode as much as possible (be explicit about what things are named, etc.).
Notation
The spec files use the following notation:
| Notation | Meaning |
|---|---|
[authenticated] |
Include access token, handle 401 refresh |
[server-only] |
Requires secretServerKey |
[BROWSER-LIKE] |
Requires browser or browser-like environment (browser, WebView, in-app browser). On mobile, open an in-app browser (ASWebAuthenticationSession on iOS, Custom Tabs on Android). On desktop, open the system browser with a registered URL scheme. |
[BROWSER-ONLY] |
Strictly requires browser environment (DOM, window object) |
[CLI-ONLY] |
Only in languages/platforms with an interactive terminal |
[JS-ONLY] |
Only available in the JavaScript SDK |
{ field, field } |
Request body (JSON) |
"Does not error" |
Function handles errors internally |
"Errors: ..." |
Lists possible errors with code/message |
See _utilities.spec.md for more details.
Language Adaptation
The languages should adapt:
- Naming conventions: camelCase (JS), snake_case (Python), PascalCase (Go), etc.
- Async patterns: Promises (JS), async/await (Python), goroutines (Go)
- Error handling: Exceptions vs Result types (language preference)
- Parameter conventions: Objects vs. kwargs, etc.
- Framework hooks: Eg. for React, add
use*equivalents toget*/list*methods - Everything else, wherever it makes sense: Every language is unique and the patterns will differ. If you have to decide between what's idiomatic in a language vs. what was done in the Hexclave SDK for other languages, use the idiomatic pattern.
Implementation Notes
Object Construction
When constructing SDK objects (User, Team, etc.) from API responses:
- Map naming conventions to your language's naming convention
- Objects should hold a reference to the SDK client for making API calls
- Objects can be mutable or immutable based on language conventions
update()methods should update local properties after successful API call
Caching
Normal functions should not cache. Some frameworks, like React, have hooks that require caching; for these, require explicit guidance.
Pagination
Most list* methods support pagination:
- Request with
cursorandlimitquery params - Response includes
pagination: { next_cursor?: string } next_cursoris null or absent when no more pages- Default limit is typically 100
- Note that not all backend APIs support pagination, and some just return all items at once.
Date/Time Formats
- API uses milliseconds since epoch for timestamps (e.g.,
signed_up_at_millis) - Convert to your language's native Date/DateTime type