feat: add list chats endpoint #1425
Conversation
🦋 Changeset detectedLatest commit: 86a54e3 The changes in this PR will be included in the next version bump. This PR includes changesets to release 3 packages
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Preview Environment (PR #1425)Preview environment scaled down after 12h of inactivity. |
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
| HAVING if( | ||
| ? = '', | ||
| true, | ||
| if( | ||
| ? = 'asc', | ||
| min(time_unix_nano) > (SELECT min(time_unix_nano) FROM telemetry_logs WHERE gram_chat_id = ? GROUP BY gram_chat_id LIMIT 1), | ||
| min(time_unix_nano) < (SELECT min(time_unix_nano) FROM telemetry_logs WHERE gram_chat_id = ? GROUP BY gram_chat_id LIMIT 1) | ||
| ) | ||
| ) | ||
| ORDER BY | ||
| IF(? = 'asc', start_time_unix_nano, 0) ASC, | ||
| IF(? = 'desc', start_time_unix_nano, 0) DESC | ||
| LIMIT ? |
There was a problem hiding this comment.
🔴 ListChats pagination is unstable and can skip/duplicate results because ORDER BY lacks a tie-breaker
ListChats paginates by comparing min(time_unix_nano) against the cursor chat's min(time_unix_nano) and orders only by start_time_unix_nano. If multiple chats share the same min(time_unix_nano) (likely with batched ingestion or coarse timestamps), pagination becomes non-deterministic:
- Page 1 might end with a chat whose
start_time_unix_nanoties with other chats. - Page 2 uses
>/<on only the timestamp, so tied chats can be skipped (never appear) or duplicated depending on ordering.
Actual: pagination may miss chats or return duplicates across pages.
Expected: deterministic pagination with a stable sort key and cursor that matches the sort.
Evidence
- Cursor predicate compares only
min(time_unix_nano):
server/internal/telemetry/repo/queries.sql.go:423-431 ORDER BYonly usesstart_time_unix_nanoand no secondary key:
server/internal/telemetry/repo/queries.sql.go:432-435
Recommendation: Add a deterministic tie-breaker to the sort and cursor comparison (e.g., (start_time_unix_nano, gram_chat_id)), and compute next_cursor based on the same composite key (or make cursor a composite token).
Was this helpful? React with 👍 or 👎 to provide feedback.
| FROM telemetry_logs | ||
| WHERE gram_project_id = ? | ||
| AND time_unix_nano >= ? | ||
| AND time_unix_nano <= ? | ||
| AND gram_chat_id IS NOT NULL | ||
| AND gram_chat_id != '' | ||
| AND (? = '' OR gram_deployment_id = toUUIDOrNull(?)) | ||
| AND if(? = '', true, position(telemetry_logs.gram_urn, ?) > 0) | ||
| GROUP BY gram_chat_id | ||
| HAVING if( | ||
| ? = '', | ||
| true, | ||
| if( | ||
| ? = 'asc', | ||
| min(time_unix_nano) > (SELECT min(time_unix_nano) FROM telemetry_logs WHERE gram_chat_id = ? GROUP BY gram_chat_id LIMIT 1), | ||
| min(time_unix_nano) < (SELECT min(time_unix_nano) FROM telemetry_logs WHERE gram_chat_id = ? GROUP BY gram_chat_id LIMIT 1) | ||
| ) | ||
| ) |
There was a problem hiding this comment.
🟡 Chat cursor subquery is not scoped by project/time filters, making pagination unstable if gram_chat_id is not globally unique
The cursor boundary subquery used in ListChats looks up the cursor chat’s min(time_unix_nano) by gram_chat_id only:
SELECT min(time_unix_nano) FROM telemetry_logs WHERE gram_chat_id = ? GROUP BY gram_chat_idActual behavior: if gram_chat_id can collide across projects (or if there are logs for the same gram_chat_id outside the requested time window), the cursor’s start time may be computed from unrelated rows. This makes pagination jump backwards/forwards unexpectedly.
Expected behavior: the cursor lookup should be scoped consistently with the main query (at minimum by gram_project_id, and ideally also by the same time range / filters).
Click to expand
The main query scopes to a project/time window:
WHERE gram_project_id = ? AND time_unix_nano >= ? AND time_unix_nano <= ? ...
But the cursor subquery does not.
Recommendation: Add gram_project_id = ? (and optionally time_unix_nano bounds / same filters) to the cursor subquery so it computes the cursor boundary in the same domain as the main result set.
Was this helpful? React with 👍 or 👎 to provide feedback.
This adds an endpoint to list chats (by grouping them by chat ID).
First step into having the full view of chat traces.